<?xml version="1.0" encoding="UTF-8"?>
<xml><records><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Randomization of Data Generation Times Improves Performance of Predictive IoT Networks</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">7th IEEE World Forum on the Internet of Things</style></secondary-title><urls><style face="normal" font="default" size="100%">https://wfiot2021.iot.ieee.org</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/WF-IoT51360.2021.9595819</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Input traffic from Internet of Things (IoT) devices is often both periodic and requires to be received by a given deadline. This can create congestion at instants of time when traffic flowing from multiple devices arrives at a shared input port or gateway, resulting in missed deadlines at the receiver. As a consequence, scheduling techniques such as the “Earliest Deadline First” (EDF) and “Priority based on Average Load” (PAL) are used to schedule the flow from different devices so as to try to satisfy the needs of the largest number of traffic flows in a timely fashion. In this paper, we propose the Randomization of flow Genaration Times (RGT) in order to smooth the total incoming traffic to the input port or gateway, on top of the use of EDF and PAL. We then evaluate the performance of RGT together with PAL and EDP, for traffic load with a varying number of up to 6400 IoT devices. Our simulation results show that RGT provides significantly better performance when added to EDF and PAL. Also, the additional computation required by RGT at each device can be quite small, suggesting that RGT is a very useful addition for improving the performance of IoT networks.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">New Orleans, Louisiana, USA</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author><author><style face="normal" font="default" size="100%">Osman Yıldız</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Recurrent Trend Predictive Neural Network for Multi-Sensor Fire Detection</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year><pub-dates><date><style face="normal" font="default" size="100%">10/2021</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/abstract/document/9451553</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2021.3087736</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">9</style></volume><section><style face="normal" font="default" size="100%">84204</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We propose a Recurrent Trend Predictive Neural Network (rTPNN) for multi-sensor fire detection based on the trend as well as level prediction and fusion of sensor readings. The rTPNN model significantly differs from the existing methods due to recurrent sensor data processing employed in its architecture. rTPNN performs trend prediction and level prediction for the time series of each sensor reading and captures trends on multivariate time series data produced by multi-sensor detector. We compare the performance of the rTPNN model with that of each of the Linear Regression (LR), Nonlinear Perceptron (NP), Multi-Layer Perceptron (MLP), Kendall-τ combined with MLP, Probabilistic Bayesian Neural Network (PBNN), Long-Short Term Memory (LSTM), and Support Vector Machine (SVM) on a publicly available fire data set. Our results show that rTPNN model significantly outperforms all of the other models (with 96% accuracy) while it is the only model that achieves high True Positive and True Negative rates (both above 92%) at the same time. rTPNN also triggers an alarm in only 11 s from the start of the fire, where this duration is 22 s for the second-best model. Moreover, we present that the execution time of rTPNN is acceptable for real-time applications.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Roozbeh Qorbanian</style></author><author><style face="normal" font="default" size="100%">Deniz Eliiyi</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Multi-Channel Joint Forecasting-Scheduling for the Internet of Things</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2020</style></year><pub-dates><date><style face="normal" font="default" size="100%">11/2021</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/abstract/document/9260218</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2020.3038358</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">8</style></volume><section><style face="normal" font="default" size="100%">217324 </style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We develop a methodology for Multi-Channel Joint Forecasting-Scheduling (MC-JFS) targeted at solving the Medium Access Control (MAC) layer Massive Access Problem of Machine-to-Machine (M2M) communication in the presence of multiple channels, as found in Orthogonal Frequency Division Multiple Access (OFDMA) systems. In contrast with the existing schemes that merely react to current traffic demand, Joint Forecasting-Scheduling (JFS) forecasts the traffic generation pattern of each Internet of Things (IoT) device in the coverage area of an IoT Gateway and schedules the uplink transmissions of the IoT devices over multiple channels in advance, thus obviating contention, collision and handshaking, which are found in reactive protocols. In this paper, we present the general form of a deterministic scheduling optimization program for MC-JFS that maximizes the total number of bits that are delivered over multiple channels by the delay deadlines of the IoT applications. In order to enable real-time operation of the MC-JFS system, first, we design a heuristic, called Multi-Channel Look Ahead Priority based on Average Load (MC-LAPAL), that solves the general form of the scheduling problem. Second, for the special case of identical channels, we develop a reduction technique by virtue of which an optimal solution of the scheduling problem is computed in real time. We compare the network performance of our MC-JFS scheme against Multi-Channel Reservation-based Access Barring (MC-RAB) and Multi-Channel Enhanced Reservation-based Access Barring (MC-ERAB), both of which serve as benchmark reactive protocols. Our results show that MC-JFS outperforms both MC-RAB and MC-ERAB with respect to uplink cross-layer throughput and transmit energy consumption, and that MC-LAPAL provides high performance as an MC-JFS heuristic. Furthermore, we show that the computation time of MC-LAPAL scales approximately linearly with the number of IoT devices. This work serves as a foundation for building scalable JFS schemes at IoT Gateways in the near future.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Alperen Helva</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Subspace-Based Emulation of the Relationship Between Forecasting Error and Network Performance in Joint Forecasting-Scheduling for the Internet of Things</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">7th IEEE World Forum on the Internet of Things</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/WF-IoT51360.2021.9595663</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We develop a novel methodology that discovers the relationship between the forecasting error and the performance of the application that utilizes the forecasts. In our methodology, an Artificial Neural Network (ANN) learns this relationship while the forecasting error is kept inside a subspace of the entire space of forecasting errors during training. We apply our methodology to the case of Joint Forecasting-Scheduling (JFS) for the Internet of Things (IoT). Our results hold potential to improve the performance of JFS in next-generation networks and can be applied to a much wider range of problems beyond IoT.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">New Orleans</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Kubilay Karakayali</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">An End-to-End Trainable Feature Selection-Forecasting Architecture Targeted at the Internet of Things</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/document/9477183</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2021.3092228</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We develop a novel end-to-end trainable feature selection-forecasting (FSF) architecture for predictive networks targeted at the Internet of Things (IoT). In contrast with the existing filter-based, wrapper-based and embedded feature selection methods, our architecture enables the automatic selection of features dynamically based on feature importance score calculation and gamma-gated feature selection units that are trained jointly and end-to-end with the forecaster. We compare the performance of our FSF architecture on the problem of forecasting IoT device traffic against the following existing (feature selection, forecasting) technique pairs: Autocorrelation Function (ACF), Analysis of Variance (ANOVA), Recurrent Feature Elimination (RFE) and Ridge Regression methods for feature selection, and Linear Regression, Multi-Layer Perceptron (MLP), Long Short Term Memory (LSTM), 1 Dimensional Convolutional Neural Network (1D CNN), Autoregressive Integrated Moving Average (ARIMA), and Logistic Regression for forecasting. We show that our FSF architecture achieves either the best or close to the best performance among all of the competing techniques by virtue of its dynamic, automatic feature selection capability. In addition, we demonstrate that both the training time and the execution time of FSF are reasonable for IoT applications. This work represents a milestone for the development of predictive networks for IoT in smart cities of the near future.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erdem Çakan</style></author><author><style face="normal" font="default" size="100%">Aral Şahin</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Multi-Layer Perceptron Decomposition Architecture for Mobile IoT Indoor Positioning</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">7th IEEE World Forum on the Internet of Things</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/WF-IoT51360.2021.9595282</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We develop a Multi-Layer Perceptron (MLP) Decomposition&lt;br /&gt;&#13;
architecture for mobile Internet Things (IoT) indoor&lt;br /&gt;&#13;
positioning. We demonstrate the performance of our architecture&lt;br /&gt;&#13;
on an indoor system that utilizes ultra-wideband (UWB) positioning.&lt;br /&gt;&#13;
Our architecture outperforms the following benchmark&lt;br /&gt;&#13;
processing techniques on the same data: MLP, Linear Regression,&lt;br /&gt;&#13;
Ridge Regression, Support Vector Regression, and the Least&lt;br /&gt;&#13;
Squares Method for indoor positioning. The results show that our&lt;br /&gt;&#13;
architecture can significantly advance the positioning accuracy&lt;br /&gt;&#13;
of indoor positioning systems and enable indoor applications&lt;br /&gt;&#13;
such as navigation, proximity marketing, asset tracking, collision&lt;br /&gt;&#13;
avoidance, and social distancing.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">MIRAI Botnet Attack Detection with Auto-Associative Dense Random Neural Network</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year><pub-dates><date><style face="normal" font="default" size="100%">7-11 Dec 2021</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">2021 IEEE Global Communications Conference</style></secondary-title><abstract><style face="normal" font="default" size="100%">Internet connected IoT devices have often been particularly vulnerable to Botnet attacks of the Mirai family in recent years. Thus we develop an attack detection scheme for Mirai Botnets, using the Auto-Associative Dense Random Neural Network that has recently been successful for other attacks such as the SYN attack. The resulting method is trained with normal traffic and tested with attack traffic, and shown to result in high accuracy detection of attacks with low false alarms. The approach is compared on the same data set with two other common Machine learning methods (Lasso and KNN) and shown to have higher accuracy, and much lower computation times than KNN and slightly higher (but comparable) computation times with respect to Lasso.</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Barcelona</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Dariusz Marek</style></author><author><style face="normal" font="default" size="100%">Tadeusz Czachórski</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Diffusion Analysis Improves Scalability of IoTNetworks to Mitigate the Massive Access Problem</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year><pub-dates><date><style face="normal" font="default" size="100%">11/2021</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">2021 Mascots: 29th International Symposium on the Modelling, Analysis and Simulation of Computer and Telecommunication Systems</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/MASCOTS53633.2021.9614289</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;A significant challenge of IoT networks is to offer Quality of Service (QoS) and meet deadline requirements when packets from a massive number of IoT devices are forwarded to an IoT gateway. Many IoT devices tend to report their data to their wired or wireless network gateways at closely correlated instants of time, leading to congestion known as the MassiveAccess Problem (MAP), which increases the probability that the IoT data will not meet its required deadlines. Since IoT data loses much of its value if it arrives to destination beyond a required deadline, MAP has been extensively studied in the literature. Thus we first take a queueing theoretic view of the problem, and also use a Diffusion Approximation to gain insight into the IoT traffic statistics that affect MAP. Then we introduce the Quasi-Deterministic Transmission Policy (QDTP) which significantly alleviates MAP when the average traffic rate grows beyond a given level and substantially reduces the probability that IoT data deadlines are missed. The results are validated using real IoT data which has been placed in IP packets for transmission.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Houston, Texas, USA</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Alper Saylam</style></author><author><style face="normal" font="default" size="100%">Rifat Cikmazel</style></author><author><style face="normal" font="default" size="100%">Nur Kelesoglu</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Energy-Efficient Indoor Positioning for Mobile Internet of Things Based on Artificial Intelligence</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">ASYU2021 </style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ASYU52992.2021.9599049</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We develop an energy-efficient indoor positioning system based on Artificial Intelligence (AI). In our system, first, at the positioning layer, a Multi-Layer Perceptron (MLP) estimates the current indoor position of an IoT device based on positioning indicators obtained from the anchors. Second, at the forecasting layer, a pair of MLPs estimate the future positions of the device based on the past position estimates obtained when the device woke up as well as the forecast positions of the device during the sleep periods. Third, the device is awakened to send a positioning beacon at intervals over which a significant displacement is predicted to occur by the forecasting layer. Our results demonstrate that our indoor positioning system saves significant energy via adaptive sleep cycles whose duration is determined by the &#160;prediction of a significant displacement. This work establishes a foundation for indoor positioning that utilizes AI-based positioning and trajectory forecasting.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Onur Çopur</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Comparative Study of Forecasting Models for COVID-19 Outbreak in Turkey</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">ASYU2021 </style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ASYU52992.2021.9599053</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;This paper gives an explanation for the failure of machine learning models for the prediction of the cases and the other future trends of Covid-19 pandemic. The paper shows that simple Linear Regression models provide high prediction accuracy values reliably but only for a 2-weeks period and that relatively complex machine learning models, which have the potential of learning long-term predictions with low errors, cannot achieve to obtain good predictions with possessing a high generalization ability. It is suggested in the paper that the lack of a sufficient number of samples is the source of the low prediction performance of the forecasting models. To exploit the information, which is of most relevant with the active cases, we perform feature selection over a variety of variables such as the numbers of active cases, deaths, recoveries, and population. Furthermore, we compare Linear Regression, Multi-Layer Perceptron, and Long-Short Term Memory models each of which is used for prediction of active cases together with various feature selection methods. Our results show that the accurate forecasting of the active cases with high generalization ability is possible up to 3 days because of the small sample size of COVID-19 data. We observe that the Linear Regression model has much better prediction performance with high generalization ability as compared to the complex models but, as expected, its performance decays sharply for more than 14-days prediction horizons.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Alper Saylam</style></author><author><style face="normal" font="default" size="100%">Nur Kelesoglu</style></author><author><style face="normal" font="default" size="100%">Rifat Cikmazel</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Dynamic Positioning Interval Based On Reciprocal Forecasting Error (DPI-RFE) Algorithm for Energy-Efficient Mobile IoT Indoor Positioning</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">International Conference on Computer, Information and Telecommunication Systems (CITS)</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/CITS52676.2021.9618231</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We develop an algorithm called “Dynamic Positioning Interval based on Reciprocal Forecasting Error (DPIRFE)” for energy-efficient mobile Internet of Things (IoT) Indoor Positioning (IP). In contrast with existing IP algorithms, DPIRFE forecasts the future trajectory of a mobile IoT device by using machine learning and dynamically adjusts the positioning interval based on the reciprocal instantaneous forecasting error, thereby dynamically trading off transmit energy consumption against forecasting error. We compare the performance of DPIRFE with respect to the total transmit energy consumption and the average forecasting error against Constant Positioning Interval (CPI) and Positioning Interval based on Displacement (PID) algorithms. Our results show that DPI-RFE significantly outperforms both of these benchmark algorithms with respect to transmit energy consumption while achieving a competitive average forecasting error performance. These results open the way to the design of machine learning based trajectory forecasting algorithms that can be utilized for energy-efficient positioning in next-generation wireless networks.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Istanbul, Turkey</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Arda Asut</style></author><author><style face="normal" font="default" size="100%">Cennet Kocabıyık</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">A Smart Home Demand Response System based on Artificial Neural Networks Augmented with Constraint Satisfaction Heuristic</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2021</style></year></dates><secondary-title><style face="normal" font="default" size="100%">13th INTERNATIONAL CONFERENCE on ELECTRICAL and ELECTRONICS ENGINEERING (ELECO)</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Distributing the peak load and alleviating grid stress by considering hourly electricity prices are some of the main research problems for current smart grid systems. This paper deals with the scheduling problem of home appliances' operating hours in smart grids, which aims to achieve minimum cost in user-defined operation intervals. To this end, scheduling via Artificial Neural Networks Augmented with Constraint Satisfaction Heuristic (ANN-AH) method that emulates the operation of the optimization for smart home demand response is developed. Our results show that a home demand response via ANN-AH achieves close to optimal performance with 10 times lower execution time than the optimal scheduling. These results suggest that the ANN-AH based demand response is highly successful and practical, and it is promising for future applications in micro-grid and decentralized renewable energy systems.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Bursa, Turkey</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erdem Çakan</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Dynamic Automatic Forecaster Selection via Artificial Neural Network Based Emulation to Enable Massive Access for the Internet of Things</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year></dates><secondary-title><style face="normal" font="default" size="100%">Journal of Network and Computer Applications (JNCA)</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">https://doi.org/10.1016/j.jnca.2022.103360</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;The Massive Access Problem of the Internet of Things (IoT) occurs at the uplink Medium Access Control (MAC) layer when a massive number of IoT devices seek to transfer their data to an IoT gateway. Although recently proposed predictive access solutions that schedule the uplink traffic based on forecasts of IoT device traffic achieve high network performance, these solutions depend heavily on the performance of forecasters. Hence, the design and selection of forecasting schemes are key to enabling massive access for such predictive access solutions. To this end, in this paper, first, we develop a framework that emulates the relationship between the IoT device class composition in the coverage area of an IoT gateway and the resulting network performance by virtue of an Artificial Neural Network (ANN). Second, based on this framework, we develop the Dynamic Automatic Forecaster Selection (DAFS) method, which selects the best-performing forecasting scheme for predictive access, in particular for &#160;Joint Forecasting-Scheduling (JFS), in a manner that adapts dynamically to a changing number of IoT devices in each device class in the coverage area. We evaluate the performance of DAFS via simulations and show that our method is able to achieve at least $80\%$ of the best performance that can be attained for both throughput and energy consumption. Furthermore, we demonstrate that DAFS is robust with respect to the selection of architectural parameters and has a reasonable computation time for real-time IoT applications. These results imply that DAFS holds the potential for practical implementation at IoT gateways in order to enable massive access under a dynamically changing composition of IoT devices.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Onur Çopur</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Simone Scardapane</style></author><author><style face="normal" font="default" size="100%">Jürgen Slowack</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Engagement Detection with Multi-Task Training in E-Learning Environments</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year></dates><secondary-title><style face="normal" font="default" size="100%">International Conference on Image Analysis and Processing (ICIAP)</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1007/978-3-031-06433-3_35</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Recognition of user interaction, in particular engagement detection, became highly crucial for online working and learning environments, especially during the COVID-19 outbreak. Such recognition and detection systems significantly improve the user experience and efficiency by providing valuable feedback. In this paper, we propose a novel Engagement Detection with Multi-Task Training (ED-MTT) system which minimizes mean squared error and triplet loss together to determine the engagement level of students in an e-learning environment. The performance of this system is evaluated and compared against the state-ofthe-art on a publicly available dataset as well as videos collected from real-life scenarios. The results show that ED-MTT achieves 6% lower MSE than the best state-of-the-art performance with highly acceptable training time and lightweight feature extraction.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">Springer</style></publisher><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Dariusz Marek</style></author><author><style face="normal" font="default" size="100%">Tadeusz Czachórski</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Mitigating the Massive Access Problem in the Internet of Things</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year></dates><secondary-title><style face="normal" font="default" size="100%"> EuroCybersec</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1007/978-3-031-09357-9_10</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;The traffic from the large number of IoT devices connected to the IoT is a source of congestion known as the Massive Access Problem (MAP), that results in packet losses, delays and missed deadlines for realtime data. This paper reviews the literature on MAP and summarizes recent results on two approaches that have been designed to mitigate MAP. One approach is based on randomizing the packet arrival instants to IoT gateways within a given time interval that is chosen so that packet arrivals do not exceed their deadlines, but also so that they do not constitute bulk arrivals. The second approach is a novel traffic shaping policy named the Quasi-Deterministic-Transmission-Policy (QDTP) which has been proved to drastically reduce queue formation at the receiving gateway by delaying packet departures from the IoT devices in a judicious manner. Both analytical and experimental results are summarized, that describe the benefits of these techniques.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">Springer</style></publisher><pub-location><style face="normal" font="default" size="100%">Nice, France</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Botnet Attack Detection with Incremental Online Learning</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year></dates><secondary-title><style face="normal" font="default" size="100%">EuroCybersec 2021</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1007/978-3-031-09357-9_5</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;In recent years, IoT devices have often been the target of Mirai Botnet attacks. This paper develops an intrusion detection method based on Auto-Associated Dense Random Neural Network with incremental online learning, targeting the detection of Mirai Botnet attacks. The proposed method is trained only on benign IoT traffic while the IoT network is online; therefore, it does not require any data collection on benign or attack traffic. Experimental results on a publicly available dataset have shown that the performance of this method is considerably high and very close to that of the same neural network model with offline training. In addition, both the training and execution times of the proposed method are highly acceptable for real-time attack detection.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">Springer</style></publisher><pub-location><style face="normal" font="default" size="100%">Nice, France</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Miltiadis Siavvas</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Dimitrios Tsoukalas</style></author><author><style face="normal" font="default" size="100%">Ilias Kalouptsoglou</style></author><author><style face="normal" font="default" size="100%">Maria Mathioudaki</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Dionysios Kehagias</style></author><author><style face="normal" font="default" size="100%">Dimitrios Tzovaras</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">The IoTAC Software Security-by-Design Platform: Concept, Challenges, and Preliminary Overview</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year></dates><secondary-title><style face="normal" font="default" size="100%">18th International Conference on the Design of Reliable Communication Networks (DRCN)</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/DRCN53993.2022.9758028</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Critical everyday activities handled by modern IoT Systems imply that security is of major concern both for the end-users and the industry. Securing the IoT System Architecture is commonly used to strengthen its resilience to malicious attacks. However, the security of software running on the IoT must be considered as well, since the exploitation of its vulnerabilities can infringe the security of the overall system, regardless of how secure its architecture may be. Thus, we present an IoT Software Security-by-Design (SSD) Platform, which provides mechanisms for monitoring and optimizing the security of IoT software applications throughout their development lifecycle, to validate the broader security of the IoT software. This paper describes the proposed SSD platform that leverages security information from all phases of development, using some novel mechanisms that have been implemented, and which can lead to a holistic security evaluation and future security certification.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Vilanova i la Geltrú, Spain</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Baran Gül</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Predictability of Internet of Things Traffic at the Medium Access Control Layer Against Information-Theoretic Bounds</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year><pub-dates><date><style face="normal" font="default" size="100%">05/2022</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2022.3174126</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">10</style></volume><section><style face="normal" font="default" size="100%">55602</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Most of the existing Medium Access Control (MAC) layer protocols for the Internet of Things (IoT) model the traffic generated by each IoT device via random arrivals such as those in a Poisson process. Under this model, since it is implied that IoT device traffic cannot be predicted, only reactive MAC-layer protocols in which the network responds to the current traffic are viable. In contrast, recent work has demonstrated that the traffic generated by an individual IoT device can be predictable, thus enabling predictive network protocols at the MAC layer. In this paper, we investigate information-theoretic bounds on the predictability of IoT traffic of individual devices. To this end, first, we compare the performance achieved by the following state-of-the-art forecasters on individual IoT device traffic: Logistic Regression, Multi-Layer Perceptron (MLP), 1-Dimensional Convolutional Neural Network (1D CNN), and Long Short Term Memory (LSTM) as well as MLP under feature selection based on Analysis of Variance (ANOVA) and Auto-Correlation Function (ACF). Second, we quantify the gap between the performance of these forecasters against information-theoretic bounds as follows: For IoT devices that generate a fixed number of bits at each generation instance, we measure the gap between the forecasting accuracy and the information-theoretic bound established by Fano’s inequality on the probability of correct prediction. Our empirical results show that existing forecasting schemes perform close to the information-theoretic bound in this case. For IoT devices that generate a variable number of bits, we measure the gap between the Mean Square Error (MSE) and the estimation-theoretic counterpart to Fano’s inequality. Our empirical results show that the performance of existing forecasting schemes is far from the information-theoretic bound in this case. This work motivates the machine learning community to develop forecasting schemes that approach information-theoretic bounds. Furthermore, this work is expected to impact the development of predictive MAC-layer protocols that exploit these bounds.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Tadeusz Czachórski</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Improving Massive Access to IoT Gateways</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year><pub-dates><date><style face="normal" font="default" size="100%">08/2022</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">Performance Evaluation</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1016/j.peva.2022.102308</style></electronic-resource-num><section><style face="normal" font="default" size="100%">102308</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;IoT networks handle incoming packets from large numbers of IoT Devices (IoTDs) to IoT Gateways. This can lead to the IoT Massive Access Problem that causes buffer overflow, large end-to-end delays and missed deadlines. This paper analyzes a novel traffic shaping method named the Quasi-Deterministic Traffic Policy (QDTP) that mitigates this problem by shaping the incoming traffic without increasing the end-to-end delay or dropping packets. Using queueing theoretic techniques and extensive data driven simulations with real IoT datasets, the value of QDTP is shown as a means to considerably reduce congestion at the Gateway, and significantly improve the IoT network’s overall performance.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Arif Dayı</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Buse Pehlivan</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Multi-Channel Subset Iteration with Minimal Loss in Available Capacity (MC-SIMLAC) Algorithm for Joint Forecasting-Scheduling in the Internet of Things</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year><pub-dates><date><style face="normal" font="default" size="100%">06/2022</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">Journal of Wireless Mobile Networks, Ubiquitous Computing, and Dependable Applications (JoWUA)</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.22667/JOWUA.2022.06.30.068</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">13</style></volume><issue><style face="normal" font="default" size="100%">2</style></issue><section><style face="normal" font="default" size="100%">68</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;The Massive Access Problem of the Internet of Things (IoT) refers to the problem of scheduling the uplink transmissions of a massive number of IoT devices in the coverage area of an IoT gateway. Joint Forecasting-Scheduling (JFS) is a recently developed methodology in which an IoT gateway forms predictions of the future uplink traffic generation pattern of each IoT device in its coverage area via machine learning algorithms and uses these predictions to schedule the uplink traffic of all of the IoT devices in advance. In this paper, we develop a novel algorithm, which we call “Multi-Channel Subset Iteration with Minimal Loss in Available Capacity” (MC-SIMLAC), for multi-channel joint forecasting-scheduling. Our multi-channel scheduling algorithm iterates over subsets of all of the bursts of IoT device traffic and selects channel-slot pairs by targeting the minimization of loss in total available capacity. In this regard, our algorithm contrasts sharply with Multi-Channel Look Ahead Priority based on Average Load (MC-LAPAL), which is the best-performing heuristic that has been developed so far for multi-channel JFS. In the general case, our algorithm outperforms MC-LAPAL, especially when wireless links operate in the power-limited regime and the number of devices is large. For the special case of identical channels, our algorithm achieves a performance that is closer than MC-LAPAL to that of the optimal scheduler. Furthermore, we prove that the time complexity and the space complexity of MC-SIMLAC in the worst case are polynomial in each of the system parameters, which indicates practical feasibility. These results pave the way to the widespread use of multi-channel joint forecasting-scheduling at IoT gateways in the near future.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">G-Networks Can Detect Different Types of Cyberattacks</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year></dates><secondary-title><style face="normal" font="default" size="100%">2022 Mascots: 30th International Symposium on the Modelling, Analysis and Simulation of Computer and Telecommunication Systems</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Malicious network attacks are a serious source of concern, and machine learning techniques are widely used to build Attack Detectors with off-line training with real attack and non-attack data, and used online to monitor system entry points connected to networks. Many machine learning based Attack Detectors are typically trained to identify specific types attacks, and the training of such algorithms to cover several types of attacks may be excessively time consuming. This paper shows that G-Networks, which are queueing networks with product form solution and special customers such as negative customers and triggers, can be trained just with “non-attack” traffic, can accurately detect several different attack types. This is established with a special case of G-Networks with triggerred customer movement. A DARPA attack and non-attack traffic repository is used to train and test the the G-Network, yielding comparable or clearly better accuracy than most known attack detection techniques.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Nice, France</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Alperen Helva</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author><author><style face="normal" font="default" size="100%">Volkan Rodoplu</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">MOSAL: A Subspace Based Forecasting Algorithm for Throughput Maximization in IoT Networks</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year><pub-dates><date><style face="normal" font="default" size="100%">11/2022</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Sensors Journal</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/document/9942929</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/JSEN.2022.3219251</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Predictive solution techniques have been developed recently in order to solve the Massive Access Problem of the Internet of Things (IoT). These techniques forecast the traffic generation patterns of individual IoT devices in the coverage area of an IoT gateway and schedule the MAC-layer resources at the gateway in advance based on these forecasts. Although predictive solutions have achieved high network performance, a key problem is that their performance depends highly on the performance of forecasters. In this paper, in order to minimize the effects of forecasting errors on the performance of predictive networks, we develop a subspace based forecasting algorithm called “Motion On a Subspace under Adaptive Learning rate (MOSAL)”. First, our algorithm trains a forecaster by minimizing the performance loss of an IoT network based on the emulation of an Application-Specific Error Function by an Artificial Neural Network. Second, the algorithm moves close to a subspace of the forecasting errors while it aims to maximize network throughput. Our results show that MOSAL achieves a throughput performance that surpasses the performance of commonly used standard gradient descent training algorithms at a reasonable execution time. These results open the way to the deployment of predictive solutions at IoT Gateways in practice in the near future.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Traffic Based Sequential Learning During Botnet Attacks to Identify Compromised IoT Devices</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2022</style></year><pub-dates><date><style face="normal" font="default" size="100%">10/2022</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/document/9969594</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2022.3226700</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">10</style></volume><section><style face="normal" font="default" size="100%">126536</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;A novel online Compromised Device Identification System (CDIS) is presented to identify IoT devices and/or IP addresses that are compromised by a Botnet attack, within a set of sources and destinations that transmit packets. The method uses specific metrics that are selected for this purpose and which are easily extracted from network traffic, and trains itself online during normal operation with an Auto-Associative Dense Random Neural Network (AADRNN) using traffic metrics measured as traffic arrives. As it operates, the AADRNN is trained with auto-associative learning only using traffic that it estimates as being benign, without prior collection of different attack data. The experimental evaluation on publicly available Mirai Botnet attack data shows that CDIS achieves high performance with Balanced Accuracy of 97%, despite its low on-line training and execution time. Experimental comparisons show that the AADRNN with sequential (online) auto-associative learning, provides the best performance among six different state-of-the-art machine learning models. Thus CDIS can provide crucial effective information to prevent the spread of Botnet attacks in IoT networks having multiple devices and IP addresses.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Real-Time Cyberattack Detection with Offline and Online Learning</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2023</style></year></dates><secondary-title><style face="normal" font="default" size="100%">IEEE International Symposium on Local and Metropolitan Area Networks</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/document/10189812</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/LANMAN58293.2023.10189812</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;This paper presents several novel algorithms for real-time cyberattack detection using the Auto-Associative Deep Random Neural Network, which were developed in the HORIZON 2020 IoTAC Project. Some of these algorithms require offline learning, while others require the algorithm to learn during its normal operation while it is also testing the flow of incoming traffic to detect possible attacks. Most of the methods we present are designed to be used at a single node, while one specific method collects data from multiple network ports to detect and monitor the spread of a Botnet. The evaluation of the accuracy of all the methods is carried out with real attack traces. These novel methods are also compared with other state-of-the-art approaches, showing that they offer better or equal performance,&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">London, United Kingdom</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Onur Çopur</style></author><author><style face="normal" font="default" size="100%">Emrah Biyik</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Renewable energy management in smart home environment via forecast embedded scheduling based on Recurrent Trend Predictive Neural Network</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2023</style></year><pub-dates><date><style face="normal" font="default" size="100%">03/2023</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">Applied Energy</style></secondary-title><urls><style face="normal" font="default" size="100%">https://authors.elsevier.com/a/1gr4J15eif8Fs5</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1016/j.apenergy.2023.121014</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">340</style></volume><section><style face="normal" font="default" size="100%">121014</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Smart home energy management systems help the distribution grid operate more efficiently and reliably, and enable effective penetration of distributed renewable energy sources. These systems rely on robust forecasting, optimization, and control/scheduling algorithms that can handle the uncertain nature of demand and renewable generation. This paper proposes an advanced ML algorithm, called Recurrent Trend Predictive Neural Network based Forecast Embedded Scheduling (rTPNN-FES), to provide efficient residential demand control. rTPNN-FES is a novel neural network architecture that simultaneously forecasts renewable energy generation and schedules household appliances. By its embedded structure, rTPNN-FES eliminates the utilization of separate algorithms for forecasting and scheduling and generates a schedule that is robust against forecasting errors. This paper also evaluates the performance of the proposed algorithm for an IoT-enabled smart home. The evaluation results reveal that rTPNN-FES provides near-optimal scheduling 37.5 times faster than the optimization while outperforming state-of-the-art forecasting techniques.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mohammed Nasereddin</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Measurement Based Evaluation and Mitigation of Flood Attacks on a LAN Test-Bed</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2023</style></year></dates><secondary-title><style face="normal" font="default" size="100%">The 48th IEEE Conference on Local Computer Networks</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/document/10223358</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">https://doi.org/10.1109/LCN58197.2023.10223358</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;The IoT is vulnerable to network attacks, and Intrusion Detection Systems (IDS) can provide high attack detection accuracy and are easily installed in IoT Servers. However, IDS are seldom evaluated in operational conditions which are seriously impaired by attack overload. Thus a Local Area Network testbed is used to evaluate the impact of UDP Flood Attacks on an IoT Server, whose first line of defence is an accurate IDS. We show that attacks overload the multi-core Server and paralyze its IDS. Thus a mitigation scheme that detects attacks rapidly, and drops packets within milli-seconds after the attack begins, is proposed and experimentally evaluated.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Daytona Beach, Florida, USA</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Baran Gül</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Decentralized Online Federated G-Network Learning for Lightweight Intrusion Detection</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year></dates><secondary-title><style face="normal" font="default" size="100%">Modeling, Analysis, and Simulation On Computer and Telecommunication Systems (MASCOTS)</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/MASCOTS59514.2023.10387644</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Cyberattacks are increasingly threatening networked systems, often with the emergence of new types of unknown (zero-day) attacks and the rise of vulnerable devices. While Machine Learning (ML)-based Intrusion Detection Systems (IDSs) have been shown to be extremely promising in detecting these attacks, the need to learn large amounts of labeled data often limits the applicability of ML-based IDSs to cybersystems that only have access to private local data. To address this issue, this paper proposes a novel Decentralized and Online Federated Learning Intrusion Detection (DOF-ID) architecture. DOF-ID is a collaborative learning system that allows each IDS used for a cybersystem to learn from experience gained in other cybersystems in addition to its own local data without violating the data privacy of other systems. As the performance evaluation results using public Kitsune and Bot-IoT datasets show, DOF-ID significantly improves the intrusion detection performance in all collaborating nodes simultaneously with acceptable computation time for online learning.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Online Self-Supervised Deep Learning for Intrusion Detection Systems</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year><pub-dates><date><style face="normal" font="default" size="100%">05/2024 </style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Transactions on Information Forensics and Security</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/abstract/document/10531267</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/TIFS.2024.3402148</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">19</style></volume><section><style face="normal" font="default" size="100%">5668</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;This paper proposes a novel Self-Supervised Intrusion Detection (SSID) framework, which enables a fully online {Deep Learning (DL)} based Intrusion Detection System (IDS) that requires no human intervention or prior off-line learning. The proposed framework analyzes and labels incoming traffic packets based only on the decisions of the IDS itself using an Auto-Associative Deep Random Neural Network, and on an online estimate of its statistically measured trustworthiness. The SSID framework enables IDS to adapt rapidly to time-varying characteristics of the network traffic, and eliminates the need for offline data collection. This approach avoids human errors in data labeling, and human labor and computational costs of model training and data collection. The approach is experimentally evaluated on public datasets and compared with well-known {machine learning and deep learning} models, showing that this SSID framework is very useful and advantageous as an accurate and online learning DL-based IDS for IoT systems.&lt;/p&gt;&#13;
&#13;
&lt;p&gt;This paper proposes a novel Self-Supervised Intrusion Detection (SSID) framework, which enables a fully online Deep Learning (DL) based Intrusion Detection System (IDS) that requires no human intervention or prior off-line learning. The proposed framework analyzes and labels incoming traffic packets based only on the decisions of the IDS itself using an Auto-Associative Deep Random Neural Network, and on an online estimate of its statistically measured trustworthiness. The SSID framework enables IDS to adapt rapidly to time-varying characteristics of the network traffic, and eliminates the need for offline data collection. This approach avoids human errors in data labeling, and human labor and computational costs of model training and data collection. The approach is experimentally evaluated on public datasets and compared with well-known machine learning and deep learning models, showing that this SSID framework is very useful and advantageous as an accurate and online learning DL-based IDS for IoT systems.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">IoT Network Cybersecurity Assessment with the Associated Random Neural Network</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2023</style></year></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2023.3297977</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;This paper proposes a method to assess the security of an n device, or IP address, IoT network by simultaneously identifying all the compromised IoT devices and IP addresses. It uses a specific Random Neural Network (RNN) architecture composed of two mutually interconnected sub-networks that complement each other in a recurrent structure, called the Associated RNN (ARNN). For each of the n devices or IP addresses in the IoT network, two distinct neurons of the ARNN advocate opposite views: compromised or not compromised. The fully interconnected 2 n neuron ARNN structure of paired neurons learns offline from ground truth data. Thus rather than requiring a separate attack detector at each network node, the ARNN offers a single overall attack detector that observes the incoming traffic at each node, learns about the interdependencies between network nodes, and formulates a recommendation for each device or IP address in an IoT network. The ARNN weight initialization and learning algorithm are discussed, and the ARNN performance is evaluated using real attack data, and compared against several learning and testing techniques. Results are obtained both for off-line learning with ground truth data, and for on-line incremental learning using a simplified average metric measured from incoming packet traffic. Comparisons with the best state-of-the-art techniques show that the ARNN significantly outperforms previously known approaches.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Pervin Bulucu</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Multi-Sensor E-Nose based on Online Transfer Learning Trend Predictive Neural Network</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2024.3401569</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Electronic Nose (E-Nose) systems, widely applied across diverse fields, have revolutionized quality control, disease diagnostics, and environmental management through their odor detection and analysis capabilities. The decision and analysis of E-Nose systems often enabled by Machine Learning (ML) models that are trained offline using existing datasets. However, despite their potential, offline training efforts often prove intensive and may still fall short in achieving high generalization ability and specialization for considered application. To address these challenges, this paper introduces the e-rTPNN decision system, which leverages the Recurrent Trend Predictive Neural Network (rTPNN) combined with online transfer learning. The recurrent architecture of the e-rTPNN system effectively captures temporal dependencies and hidden sequential patterns within E-Nose sensor data, enabling accurate estimation of trends and levels. Notably, the system demonstrates the ability to adapt quickly to new data during online operation, requiring only a small offline dataset for initial learning. We evaluate the performance of the e-rTPNN decision system in two domains: beverage quality assessment and medical diagnosis, using publicly available wine quality and Chronic Obstructive Pulmonary Disease (COPD) datasets, respectively. Our evaluation indicates that the proposed e-rTPNN achieves decision accuracy exceeding 97% while maintaining low execution times. Furthermore, comparative analysis against established Machine Learning (ML) models reveals that the e-rTPNN decision system consistently outperforms these models by a significant margin in terms of accuracy.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Nur Kelesoglu</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Fire Detection and Risk Assessment via Support Vector Regression with Flattening-Samples Based Augmented Regularization</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year><pub-dates><date><style face="normal" font="default" size="100%">10/2024</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">Applied Soft Computing</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1016/j.asoc.2024.112023</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">164</style></volume><section><style face="normal" font="default" size="100%">112023</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;We propose a Hybrid Support Vector Regression (SVR) with Flattening-Samples Based Augmented Regularization (Hybrid FSR-SVR) architecture for multi-sensor fire detection and forest fire risk assessment. The Hybrid FSR-SVR is a lightweight architecture built upon the novel Flattening-Samples Based Augmented Regularization (FSR) approach and temporal trends of environmental variables. The FSR approach augments&#160;l2&#160;norm based smoothing term into an&#160;l1-l2&#160;combination, facilitating the integration of&#160;l1&#160;regularization into the SVR method, thereby enhancing generalization with minimal computational load. We evaluate the performance of Hybrid FSR-SVR using two distinct datasets covering indoor and forest fires, benchmarking against 15 machine learning models, including state-of-the-art techniques, such as Recurrent Trend Predictive Neural Network (rTPNN), Long-Short Term Memory (LSTM), Multi-Layer Perceptron (MLP), Gated Recurrent Unit (GRU), and Gradient Boosting. Our findings demonstrate that Hybrid FSR-SVR effectively assesses the risk of forest fire, enabling early preventive measures. Notably, it achieves a remarkable accuracy of 0.95 for forest fire detection and ranks third with 0.88 accuracy for indoor fire detection. Importantly, it exhibits computation times significantly lower – by 1 to 2 orders of magnitude – than the majority of compared techniques. The superior generalization ability of Hybrid FSR-SVR, facilitated by flattening-samples based augmented regularization, allows for high detection performance even with smaller training sets.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Miltiadis Siavvas</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">System-wide vulnerability of multi-component software</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year><pub-dates><date><style face="normal" font="default" size="100%">10/2024</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%"/></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1016/j.cie.2024.110453</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">196</style></volume><section><style face="normal" font="default" size="100%">110453</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;In software systems comprised of many interconnected components, the vulnerability of each component will affect the vulnerability of other components and of the system as a whole. Existing techniques allow the quantification of the vulnerability of individual components taken singly, but the assessment of their vulnerability when they are interconnected or interdependent remains a challenge. The present work addresses this problem with a novel System-Wide Vulnerability Assessment (SWVA) framework for interconnected software components, based on an Associated Random Neural Network (ARNN) that estimates the system-wide vulnerability of all software components from known local vulnerabilities of individual components, and from their interconnections. The ARNN uses a problem-specific weight initialization, and learns from existing software system examples with a gradient-based deep learning algorithm. The ARNN is then used to assess the vulnerability of hitherto unseen software systems. The performance of the proposed ARNN-based SWVA framework is evaluated and compared against several well-known machine learning techniques on 13 different versions of a real-world software system with up to 11 components. The experimental results show the superior performance of the ARNN achieving above 85% median accuracy and good high scalability with respect to the number of connected software components.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Baran Gül</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">DISFIDA: Distributed Self-Supervised Federated Intrusion Detection Algorithm with Online Learning for Health Internet of Things and Internet of Vehicles  </style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year><pub-dates><date><style face="normal" font="default" size="100%">12/2024</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">Internet of Things</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">https://doi.org/10.1016/j.iot.2024.101340</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">28</style></volume><section><style face="normal" font="default" size="100%">101340</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Networked health systems are often the victims of cyberattacks with serious consequences for patients and healthcare costs, with the Internet of Things (IoT) being an additional prime target. In future systems we can imagine that the Internet of Vehicles (IoV) will also be used for conveying patients for diagnosis and treatment in an integrated manner. Thus the medical field poses very significant and specific challenges since even for a single patient, several providers may carry out tests or offer healthcare services, and may have distinct interconnected sub-contractors for services such as ambulances and connected cars, connected devices or temporary staff providers, that have distinct confidentiality requirements on top of possible commercial competition. On the other hand, these distinct entities can be subject to similar or coordinated attacks, and could benefit from each others’ cybersecurity experience to better detect and mitigate cyberattacks. Thus the present work proposes a novel Distributed Self-Supervised Federated Intrusion Detection Algorithm (DISFIDA), with Online Self-Supervised Federated Learning, that uses Dense Random Neural Networks (DRNN). In DISFIDA learning data is private, and neuronal weights are shared among Federated partners. Each partner in DISFIDA combines its synaptic weights with those it receives other partners, with a preference for those weights that have closer numerical values to its own weights which it has learned on its own. DISFIDA is tested with three open-access datasets against five benchmark methods, for two relevant IoT healthcare applications: networks of devices (e.g., body sensors), and Connected Smart Vehicles (e.g., ambulances that transport patients). These tests show that the DISFIDA approach offers 100% True Positive Rate for attacks (one percentage point better than comparable state of the art methods which attain 99%) so that it does better at detecting attacks, with 99% True Negative Rate similar to state-of-the-art Federated Learning, for Distributed Denial of Service (DDoS) attacks.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mohammed Nasereddin</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Deep Learning Intrusion Detection and Mitigation of DoS Attacks</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year></dates><secondary-title><style face="normal" font="default" size="100%">32nd International Symposium on the Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (MASCOTS24)</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/abstract/document/10786560</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">https://doi.org/10.1109/MASCOTS64422.2024.10786560</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Internet of Things (IoT) networks are highly vulnerable to common network DoS and DDoS attacks, which flood limited system resources or IoT devices, overwhelming them with large numbers of attack packets. In order to mitigate such attacks, this paper develops a lightweight yet effective Intrusion Detection and Prevention System (IDPS), that sequentially detects and mitigates the attack via a Deep Random Neural Network (DRNN) and a Drop-Idle-Repeat process. The IDPS is evaluated for UDP Floods, attacks on an experimental test-bed. The results show that UDP Flood attacks can be mitigated with the proposed IDPS, allowing the system to continue routine operations, and resume communications when the attack ends.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Krakow, Poland</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Miltiadis Siavvas</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">System Wide Vulnerability and Trust in Multi-Component Communication System Software</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year><pub-dates><date><style face="normal" font="default" size="100%">09/2024</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Network </style></secondary-title><urls><style face="normal" font="default" size="100%">https://drive.google.com/file/d/1123-TES0XmyG8_l1XlmAfdWV5Jf6QPeo/view </style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/MNET.2024.3452962</style></electronic-resource-num><volume><style face="normal" font="default" size="100%">39</style></volume><issue><style face="normal" font="default" size="100%">2</style></issue><section><style face="normal" font="default" size="100%">108-114</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;In the software rich environment of 6G,&#160;systems will be surrounded by edge devices that support distributed software&#160;systems which are critical to&#160;operations. Such systems may also be subject to frequent&#160;updates or uploads of individual software components.&#160;Trust in such systems will therefore depend on our ability&#160;to rapidly ensure that such software is not vulnerable to&#160;cyberattacks or malicious compromises. Thus this paper&#160;presents a novel System-Wide Vulnerability Assessment&#160;(SWVA) framework based on Machine Learning, that&#160;can be frequently activated to assess the vulnerability of&#160;interconnected software components over edge systems.&#160;The performance of the SWVA framework is illustrated&#160;by assessing the vulnerability of 13 versions of a real world 11 component software system, and comparingthe ARNN results against the well-known ML models&#160;MLP, KNN, and Lasso. The results show the superior&#160;performance of SWVA, offering over 85% median accuracy&#160;and good scalability as the number of connected software&#160;components increases.&lt;/p&gt;&#13;
</style></abstract><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">An Associated Random Neural Network Detects Intrusions and Estimates Attack Graphs</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year><pub-dates><date><style face="normal" font="default" size="100%">12/2024</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">32nd International Conference on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS)</style></secondary-title><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/MASCOTS64422.2024.10786521</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Cyberattacks, especially Botnet Distributed Denial of Service (DDoS), increasingly target networked systems, compromise interconnected nodes by constantly spreading malware. In order to prevent these attacks in their early stages, which includes stopping the spread of malware, it is vital to identify compromised nodes and successfully predict potential attack paths. To this end, this paper proposes a novel system based on an Associated Random Neural Network (ARNN) that simultaneously detects intrusion at the network-level and estimates the network attack graph. In this system, ARNN is trained online to minimize problem-specific multi-task loss so that it identifies compromised network nodes, while the neural network connection weights also estimate the attack path. The performance of the method is calculated using the Kitsune attack dataset, showing that the method achieves a recall rate above 0.95 in estimating the network attack graph, and provides a near-perfect classification of compromised nodes. The ARNN-based system for dynamic and continuous estimation of compromised nodes and network attack graphs, can pave the way for enhancing security measures, and stopping Botnet DDoS attacks from spreading in networked systems.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Krakow, Poland</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Analysis of the QoS Prediction via Recurrent Trend Predictive Neural Network under Low-Rate DoS Attacks</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2024</style></year><pub-dates><date><style face="normal" font="default" size="100%">11/2024</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">The 10th International Conference on Next Generation Computing</style></secondary-title><urls><style face="normal" font="default" size="100%">https://www.earticle.net/Article/A468884</style></urls><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Low-Rate Denial of Service (LDoS) attacks raise an increasingly frequent and significant threat to performancecritical and sensitive networks. Due to their slowly evolving nature, it is challenging –but crucial– to detect such attacks during their early phases in order to mitigate their impact on network performance, e.g. Quality of Service (QoS), in longterm operation. To this end, this paper investigates the prediction of QoS via a modified version of the Recurrent Trend Predictive Neural Network (rTPNN) and the use of the prediction towards detracting LDoS attacks. The presented rTPNN-based QoS predictor is evaluated and compared against benchmark models for five scenarios using an open-access dataset. The results have shown that the modified rTPNN model can predict QoS with under 2% SMAPE, and the QoS prediction is a promising approach for developing LDoS attack detectors in future works.&lt;/p&gt;&#13;
</style></abstract><publisher><style face="normal" font="default" size="100%">Korean Institute of Next Generation Computing</style></publisher><pub-location><style face="normal" font="default" size="100%">Philippines (Hybrid)</style></pub-location><language><style face="normal" font="default" size="100%">eng</style></language></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Nur Kelesoglu</style></author><author><style face="normal" font="default" size="100%">Taylan Mert</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Self-Supervised Online Learning Algorithm for rTPNN to Reduce False Alarms in Fire Detection</style></title></titles><keywords><keyword><style face="normal" font="default" size="100%">Fire detection</style></keyword><keyword><style face="normal" font="default" size="100%">Self-Supervised</style></keyword><keyword><style face="normal" font="default" size="100%">online learning</style></keyword><keyword><style face="normal" font="default" size="100%">human-in-the-loop</style></keyword><keyword><style face="normal" font="default" size="100%">multi-sensor</style></keyword></keywords><dates><year><style face="normal" font="default" size="100%">2026</style></year><pub-dates><date><style face="normal" font="default" size="100%">07/2026</style></date></pub-dates></dates><secondary-title><style face="normal" font="default" size="100%">Knowledge-Based Systems</style></secondary-title><keyword><style face="normal" font="default" size="100%"/></keyword><electronic-resource-num><style face="normal" font="default" size="100%">10.1016/j.knosys.2026.116533</style></electronic-resource-num><pages><style face="normal" font="default" size="100%">14</style></pages><volume><style face="normal" font="default" size="100%">350</style></volume><work-type><style face="normal" font="default" size="100%">Research article</style></work-type></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">A predictive neural network architecture for early detection of low-rate cyberattacks</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">Knowledge-Based Systems</style></secondary-title><urls><style face="normal" font="default" size="100%">https://www.sciencedirect.com/science/article/pii/S0950705126007215</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1016/j.knosys.2026.115995</style></electronic-resource-num><section><style face="normal" font="default" size="100%">115995</style></section><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Low-Rate Denial of Service (LDoS) attacks pose a significant challenge to IoT networks due to their subtle and prolonged nature, often evading traditional intrusion detection systems. This paper presents IDQS (Intrusion Detection via QoS Prediction), a lightweight and proactive framework for early LDoS attack detection. IDQS integrates two new key components: (i) RTP-QoS, a Recurrent Trend Predictive Neural Network that learns and forecasts future Quality of Service (QoS) based on historical traffic patterns, and (ii) PDM, a Pairwise Decision Model that evaluates discrepancies between predicted and actual QoS to identify potential attacks. Evaluated on the public SDN-SlowRate-DDoS and CIC-IDS2017 datasets, IDQS respectively achieves over 79% and 91% detection accuracy across most attack scenarios with high recall and low false negatives, while maintaining an end-to-end inference time of just 0.28 seconds. The results demonstrate the effectiveness and efficiency of IDQS for real-time deployment in resource-constrained IoT environments.&lt;/p&gt;</style></abstract></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Mikołaj Macura</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Performance, Energy Efficiency, and Security of Distributed rTPNN for TinyML-Enabled Smart Homes</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">32nd International Conference on Telecommunications (ICT)</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/abstract/document/11594667</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ICT70370.2026.11594667</style></electronic-resource-num><abstract><style face="normal" font="default" size="100%">&lt;p&gt;&lt;span style="-webkit-text-stroke-width:0px;background-color:rgb(255, 255, 255);color:rgb(34, 34, 34);display:inline !important;float:none;font-family:Arial, sans-serif;font-size:13px;font-style:normal;font-variant-caps:normal;font-variant-ligatures:normal;font-weight:400;letter-spacing:normal;orphans:2;text-align:start;text-decoration-color:initial;text-decoration-style:initial;text-decoration-thickness:initial;text-indent:0px;text-transform:none;white-space:normal;widows:2;word-spacing:0px;"&gt;Tiny Machine Learning (TinyML) brings AI capabilities to low-power edge devices, enabling local data processing without cloud reliance. This paper evaluates a distributed Recurrent Trend Predictive Neural Network (rTPNN) deployed on ESP8266 and ESP32 microcontrollers within a smart home IoT network. The rTPNN achieves superior accuracy with fewer parameters than standard models while incurring negligible computational costs. Energy profiling demonstrates that constant WiFi connectivity extends battery life eight-fold compared to periodic reconnection. However, security assessments reveal that constant connectivity is highly susceptible to Denial of Service (DoS) saturation, and parameter injection can degrade accuracy by over 70%. These results substantiate the feasibility of rTPNN as a TinyML for sustained operation while highlighting the critical need for codesigned security mechanisms.&lt;/span&gt;&lt;/p&gt;</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher><pub-location><style face="normal" font="default" size="100%">Thessaloniki, Greece</style></pub-location></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>17</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Pervin Bulucu</style></author><author><style face="normal" font="default" size="100%">Osman Yıldız</style></author><author><style face="normal" font="default" size="100%">Cüneyt Güzeliş</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Trends and Multidimensional Analysis of Performance-Influencing Components in Electronic Noses: A Survey</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">IEEE Access</style></secondary-title><urls><style face="normal" font="default" size="100%">https://ieeexplore.ieee.org/abstract/document/11599553</style></urls><electronic-resource-num><style face="normal" font="default" size="100%">10.1109/ACCESS.2026.3711446</style></electronic-resource-num><pages><style face="normal" font="default" size="100%">27</style></pages><section><style face="normal" font="default" size="100%">104834</style></section><abstract><style face="normal" font="default" size="100%">&lt;div class="abstract-text row g-0" style="--bs-gutter-x:0;--bs-gutter-y:0;-webkit-text-stroke-width:0px;background-color:rgb(255, 255, 255);box-sizing:border-box;clear:both;color:rgb(51, 51, 51);display:flex;flex-wrap:wrap;font-family:"HelveticaNeue Regular", sans-serif;font-size:18px;font-style:normal;font-variant-caps:normal;font-variant-ligatures:normal;font-weight:400;letter-spacing:normal;margin:0px 0px 1em !important;orphans:2;outline:none !important;text-align:start;text-decoration-color:initial;text-decoration-style:initial;text-decoration-thickness:initial;text-indent:0px;text-transform:none;white-space:normal;widows:2;word-spacing:0px;" _ngcontent-ng-c1231153542=""&gt;&lt;div class="col-12" style="box-sizing:border-box;flex:0 0 auto;margin-top:0px;max-width:100%;outline:none !important;padding-left:0px;padding-right:0px;width:886.963px;" _ngcontent-ng-c1231153542=""&gt;&lt;div class="u-mb-1" style="box-sizing:border-box;margin-bottom:1em !important;outline:none !important;" _ngcontent-ng-c1231153542=""&gt;&lt;div class="abstract-text-content" style="box-sizing:border-box;outline:none !important;" _ngcontent-ng-c1231153542="" xplmathjax=""&gt;Electronic Nose (E-Nose) systems have emerged as powerful analytical tools engineered to mimic the mammalian olfactory system, offering rapid, non-invasive, and cost-effective alternatives to traditional chemical analysis methods. Given the vast and multifaceted nature of recent research, this paper presents a detailed and multidimensional analysis of the primary components dictating E-Nose efficacy: target analytes, gas sensors, and Machine Learning (ML) algorithms. Based on a structured review of literature published between 2019 and 2025, we quantitatively explore the interdependencies among these components to identify key trends and performance drivers. Our statistical analysis reveals a continued reliance on Metal Oxide Semiconductor (MOS) sensors due to their cost-effectiveness, alongside a significant paradigm shift from traditional statistical methods toward Deep Learning (DL) architectures, particularly Convolutional Neural Networks (CNNs) and hybrid models, which are increasingly utilized to handle high-dimensional sensor data. Furthermore, we provide a granular review of applications across major sectors, including Food and Beverage, covering quality assessment, fraud detection, and flavor characterization, as well as Healthcare, and Environmental and Agricultural Monitoring. Finally, this survey critically evaluates persistent challenges such as sensor drift, detection performance, data scarcity, and explainability, while outlining emerging research opportunities in novel nanomaterials, IoT integration, Edge Computing, and multi-modal sensing to facilitate the transition from laboratory prototypes to robust commercial applications.&lt;/div&gt;&lt;/div&gt;&lt;/div&gt;&lt;/div&gt;</style></abstract></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Baran Gül</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Nasser Jazdi</style></author><author><style face="normal" font="default" size="100%">Michael Weyrich</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">QoS-Aware Federated Learning for Multimodal In-Cabin Interaction in Smart Vehicles</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD)</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Modern smart vehicles leverage multimodal sensors, ranging from high-bandwidth vision systems to low-rate physiological monitors, to provide personalized in-cabin services. However, integrating high-fidelity multimodal fusion with collaborative training is often hindered by the heterogeneous and time-varying Quality of Service (QoS) constraints of vehicular networks. Standard Federated Learning (FL) approaches enforce rigid synchronous rounds that fail to account for these resource asymmetries, leading to safety-critical timing violations and energy exhaustion. In this paper, we propose FedQoS, a novel asynchronous, event-triggered FL framework that decouples local computation from global communication via a two-phase gating mechanism. First, we introduce a resource-aware training gate that initializes local learning only when sensing buffers and energy reserves meet safety thresholds, preventing ML tasks from compromising core vehicle mobility. Second, a QoS-aware transmission policy gates uplink updates based on an efficiency score that balances model novelty against instantaneous latency and energy costs. Locally, clients optimize an objective featuring a staleness-aware proximal term that dynamically adjusts the global anchor strength based on update age. Extensive experiments on multimodal vehicular datasets demonstrate that FedQoS achieves competitive personalized accuracy with only marginal performance loss compared to FedAvg, while substantially reducing QoS violations, cutting communication overhead by 76.7%, and lowering latency cost by 26.0%, demonstrating a highly favorable accuracy and efficiency balance for real-world vehicular deployments.&lt;/p&gt;</style></abstract></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Rafał Gibała</style></author><author><style face="normal" font="default" size="100%">Anna Grygar</style></author><author><style face="normal" font="default" size="100%">Sławomir Nowak</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Trend-based Multi-Modal Anomaly and Cyber Threat Detection with Traceable Explainability for Cloud Services</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD)</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Multi-modal real-time monitoring systems based on AI are&lt;br&gt;essential for ensuring system reliability and cyber-defense in cloud computing,&lt;br&gt;but are challenged by context isolation, early-fusion computational&lt;br&gt;overhead, and insufficient transparency. In order to address these&lt;br&gt;issues, this paper introduces T-MATE, an explainable Trend-based Multimodal&lt;br&gt;Anomaly and Threat detector engineered for robust, secure cloud&lt;br&gt;infrastructure monitoring. Rather than treating multi-modal streams&lt;br&gt;monolithically, T-MATE structurally decouples data modalities into independent&lt;br&gt;neural network heads, using a specialized Recurrent Trend&lt;br&gt;Predictive Neural Network (rTPNN) to isolate underlying trends and levels&lt;br&gt;across quantitative performance telemetry, while qualitatively parsing&lt;br&gt;textual event logs via Gemini 2.5 Flash. These components output&lt;br&gt;bounded anomaly scores and are integrated at the decision level&lt;br&gt;using an Empirical Reliability-Weighted Max Fusion operator, which&lt;br&gt;scales individual outputs to enforce a max-safety posture and eliminate&lt;br&gt;parameter-explosion liabilities. In order to demonstrate operational deployment&lt;br&gt;readiness, the framework is thoroughly evaluated via 10-fold&lt;br&gt;cross-validation and compared against standalone rTPNN, LSTM, and&lt;br&gt;MLP models alongside the baseline Avg-Fuse paradigm on the public&lt;br&gt;CloudAnoBench dataset. The results reveal that T-MATE achieves a superior&lt;br&gt;overall F1 Score of 0.88, a top threshold-invariant AUC-ROC of&lt;br&gt;0.96, an exceptional True Positive Rate of 0.96, and a processing footprint&lt;br&gt;that confirms its viability for close to real-time inference in enterprise&lt;br&gt;cloud infrastructures.&lt;/p&gt;</style></abstract></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Rafał Gibała</style></author><author><style face="normal" font="default" size="100%">Sławomir Nowak</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Optimizing Information Retrieval in RAG Systems via Proactive Hyperparameter Selection</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD)</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Information retrieval is a crucial operation for Retrieval-&lt;br&gt;Augmented Generation (RAG) systems, providing substantial gains in&lt;br&gt;factual accuracy and operational eciency. However, the performance of&lt;br&gt;a RAG system is highly sensitive to the selection of hyperparameters, optimization of which is computationally expensive and time-consuming. In&lt;br&gt;this paper, we propose a novel framework, called Proactive Hyperparameter&lt;br&gt;Estimation and Selection (ProHES), which introduces a paradigm&lt;br&gt;shift from reactive, brute-force search to an ecient, analytical approach.&lt;br&gt;The core novelty of the ProHES framework is its two-stage methodology.&lt;br&gt;First, it analytically estimates the most suitable similarity metric&lt;br&gt;for a given embedding and dataset by quantifying intrinsic embedding&lt;br&gt;properties, i.e. sparsity, magnitude meaning, feature independence, and&lt;br&gt;value spread. Then, it systematically selects the optimal number of retrieved&lt;br&gt;documents based on user preferences and system requirements.&lt;br&gt;We evaluate the performance of the ProHES framework across six different&lt;br&gt;embedding functions on the MS Marco and TriviaQA datasets.&lt;br&gt;Our results demonstrate that ProHES provides a near-optimal hyperparameter&lt;br&gt;conguration with 22- to 86-fold reduction in computation&lt;br&gt;time compared to exhaustive search, achieving an average of 94% of the&lt;br&gt;best possible performance. The proposed framework represents a significant&lt;br&gt;advancement by making hyperparameter tuning for RAG systems&lt;br&gt;more ecient and systematic, thereby paving the way for more accurate,&lt;br&gt;robust and scalable real-world applications.&lt;/p&gt;</style></abstract></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Erol Gelenbe</style></author><author><style face="normal" font="default" size="100%">Baran Gül</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Federated Intrusion Detection for Smart Vehicles</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">IEEE International Conference on Intelligent Transportation Systems (ITSC)</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Cybersecurity threats endanger networked systems consisting of multiple nodes, such as the Internet of Things (IoT), Supply Chains and Smart Cars, due to the emergence of new types of unknown (zero-day) attacks against vulnerable IoT devices. While earlier research has shown that Machine Learning (ML)-based Intrusion Detection Systems (IDSs) enhance security, the efficacy and practicality of ML-based IDSs often require large training datasets based on private data. This challenge becomes more pronounced in the case of a distributed system, where each node must ensure the confidentiality of local data, yet faces attacks that are similar to those faced by other nodes in the system. Thus, this paper proposes a novel Decentralized Asynchronous IDS with Online Self-Supervised Federated Learning called the iDAF system, that improves overall security by providing collaborative confidentiality-preserving learning among all nodes. iDAF enables self-supervised learning without human intervention or large labelled data, and avoids time-consuming local training sessions. iDAF’s performance is evaluated for DoS and DDoS attacks in two different scenarios of Collaborating IoT Networks and Connected Smart Cars using three online datasets, and compared with five benchmark methods. The results show high intrusion detection accuracy with an acceptable computation time and low communication overhead.&lt;/p&gt;</style></abstract><publisher><style face="normal" font="default" size="100%">IEEE</style></publisher></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Rafał Gibała</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">An LLM-Guided Agentic Optimization Toolbox for Edge-Deployable Neural Networks</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">35th ACM International Conference on Information and Knowledge Management</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;&lt;span style="-webkit-text-stroke-width:0px;background-color:rgb(255, 255, 255);color:rgb(0, 0, 0);display:inline !important;float:none;font-family:"Helvetica Neue", Helvetica, Arial, sans-serif;font-size:12px;font-style:normal;font-variant-caps:normal;font-variant-ligatures:normal;font-weight:400;letter-spacing:normal;orphans:2;text-align:left;text-decoration-color:initial;text-decoration-style:initial;text-decoration-thickness:initial;text-indent:0px;text-transform:none;white-space:normal;widows:2;word-spacing:0px;"&gt;Deploying neural networks on resource-constrained edge devices is limited by severe envelopes on storage, memory, compute, and energy. Manual compression through pruning, weight clustering, and quantization can reduce footprint substantially, yet selecting an appropriate strategy and its hyperparameters remains a search-heavy process. In this paper, we present an automated Edge AI Optimization Toolbox that closes the loop between weight-distribution profiling, user-defined multi-objective preferences, and Large Language Model (LLM)-guided strategy synthesis for Neural Networks targeting edge export. Given a pretrained model, a calibration and evaluation dataset, and constraints on minimum accuracy together with maximum model size and RAM, the toolbox profiles layer-wise weight statistics, proposes a compression strategy among pruning, clustering, or their combination, always applies int8 quantization, re-evaluates the candidate, and iterates until the preferences are met or a maximum iteration budget is exhausted, returning either an ideal or closest-matching model. The approach is aligned with Green AI goals of reducing unnecessary computational and memory footprints at deployment time. On occupancy estimation models spanning MLP, CNN, LSTM, and public pretrained architectures, the toolbox reduces model size by roughly 72%-92% while largely preserving predictive quality. Inference latency significantly improves for several models, whereas energy and carbon impacts remain architecture-dependent under the current operator scope. These results show that preference-constrained, LLM-guided compression is an effective and promising path toward edge-deployable and resource-aware neural networks.&lt;/span&gt;&lt;/p&gt;</style></abstract></record><record><source-app name="Bibcite" version="8.x">Drupal-Bibcite</source-app><ref-type>47</ref-type><contributors><authors><author><style face="normal" font="default" size="100%">Baran Gül</style></author><author><style face="normal" font="default" size="100%">Mert Nakip</style></author><author><style face="normal" font="default" size="100%">Nasser Jazdi</style></author><author><style face="normal" font="default" size="100%">Michael Weyrich</style></author></authors></contributors><titles><title><style face="normal" font="default" size="100%">Energy- and QoS-Gated Federated Learning for Sustainable Wearable Stress Detection</style></title></titles><keywords/><dates><year><style face="normal" font="default" size="100%">2026</style></year></dates><secondary-title><style face="normal" font="default" size="100%">35th ACM International Conference on Information and Knowledge Management</style></secondary-title><abstract><style face="normal" font="default" size="100%">&lt;p&gt;Federated Learning (FL) is increasingly deployed on battery-powered edge and wearable devices, where every communication round and local training step carries a direct energy and carbon cost. Most FL evaluations, however, still report predictive accuracy alone, leaving the environmental footprint of the learning process itself unmeasured. This paper presents a two-phase, QoS- and energy-gated FL approach, FedEQ, which restricts local training to moments when a client's sensing buffer and residual energy meet safety thresholds, and restricts transmission to updates whose novelty justifies their instantaneous latency and energy cost. We evaluate FedEQ on a federated stress-detection task built from the public WESAD dataset, benchmarking it against three established aggregation schemes: FedAvg, FedProx, and the adaptive server-side optimizer FedAdam. Beyond accuracy and F1, we report communication volume, wall-clock-measured compute energy (via CodeCarbon), analytically modeled communication energy, CO2eq emissions, and floating-point operations, averaged over five random seeds. FedAdam attains the highest raw accuracy of the four methods (93.2%) at the shortest horizon we test (20 rounds); FedEQ retains 97.0% of that accuracy at the same horizon while cutting communication volume by 63.9%, total energy and CO2eq by 68.9-69.7%, and FLOPs by 78.6% relative to all three baselines, for 2.7-3.2x higher accuracy-per-resource-unit (per MB, per Wh, and per gCO2eq) than the strongest competing method. We further show this accuracy gap is a short-horizon artifact rather than a fixed property of the method: extending training to 60 rounds closes it to within 0.5 points of the best baseline - and FedEQ overtakes FedAdam outright - while its communication savings simultaneously grow to 69.1%. Across the full range tested, FedEQ is the Pareto-dominant choice among the four once predictive performance and resource cost are considered jointly.&lt;/p&gt;</style></abstract></record></records></xml>
