• Developed RAID-KL, a teacher-student framework tailored for IoT security. • Validated RAID-KL on multiple data, achieving superior accuracy and efficiency. • RAID-KL achieves 11.3% and 64.33% reduction in CPU and memory usage, respectively. • RAID-KL compressed teacher model by 91.24% with an accuracy of approx. 99.75% • Applied SHAP to interpret feature contributions for IoT attack detection. Knowledge distillation (KD) is a pivotal model compression technique that enables the deployment of lightweight neural networks on resource-constrained Internet of Things (IoT) devices without sacrificing predictive performance. In the literature, KD has been widely applied to intrusion detection designs using the Kullback-Leibler (KL) divergence loss to align the distillation loss. However, KL divergence suffers from asymmetry and numerical instability when student predictions poorly match teacher distributions. The Jensen-Shannon (JS) divergence presents an alternative that allows both distributions to be treated equally relative to their average and ensures more balanced knowledge transfer process. The JS although symmetric and bounded, may converge more slowly due to its conservative nature. This study introduces RAID-KL , a resource-aware security model that leverages KD and a novel adaptive loss function combining hybrid KL and JS divergence. By integrating the benefits of both divergences, our method enhances the generalization and convergence of distilled models while maintaining low computational overhead, balanced knowledge sharing and improved numerical robustness. RAID-KL utilizes 1D Convolutional Neural Networks (1DCNNs) as the learning algorithm with teacher-student paradigm where a complex model serves as the teacher model, transferring its learned representations to a significantly smaller student network. RAID-KL is trained and evaluated in real-world network traffic datasets, including CICIoT2023, CICIoMT2024 and NIMSLABIoT2025, which include several IoT threats. RAID-KL demonstrates high performance in all applied metrics and low resource utilization during training and inference. To elucidate the critical relationships and feature contributions in model decisions, we integrate SHapley Additive exPlanations (SHAP) values for interpretability. Our findings reveal that the choice of loss function significantly impacts both performance and resource efficiency, with the hybrid KL-JS loss achieving superior trade-offs. Empirically, RAID-KL achieves 11.3% reduction in CPU usage and 64.33% reduction in memory usage during inference. Additionally, the RAID-KL model achieves 91.24% model compression over the teacher model while maintaining nearly the same classification accuracy. These insights highlight the robustness of RAID-KL framework, while providing a nuanced explainable index compared to the literature.

This innovative practice full paper describes the challenge of engaging students in computer network education, where conveying complex concepts can be difficult. The proposed solution, P4Checkers, integrates the game of checkers with P4 programming to teach networking concepts interactively. Using the P4Docker simulation environment, the game transforms network packets into game actions, providing hands-on experience with network protocols. The effectiveness of P4Checkers was evaluated using the Post-Study System Usability Questionnaire (PSSUQ), showing positive results in improving student engagement and understanding of network concepts. This study highlights the potential of serious games like P4Checkers to enhance network education.

IPv6 Routing Protocol for Low-Power and Lossy Networks (RPL) is the only standardized routing protocol for Low Power and Lossy Networks. Therefore, RPL is the routing protocol chosen for most Internet of Things (IoT) applications. However, the RPL protocol has some open issues when the scenario includes mobile nodes. Since many IoT applications include mobility, this is an important problem to address. Node mobility does not permit the network to stabilize and because of this keep exchanging control messages. Mobility increases the packet loss because mobile nodes may leave the coverage area before transmitting the data. This problems also increase the power consumption. Another important factor affects the RPL performance. To save nodes energy, the data link layer implements a radio duty cycle. These duty cycles turn the radio off when it is not used. One common and used radio duty cycle protocol is the ContikiMAC. Changes in the frequency of the duty cycle modifies the RPL performance. This work shows how the change of radio duty cycle of ContikiMAC on a hybrid scenario, composed by fixed and mobile nodes, interferes on RPL performance. Results show that the nodes mobility degrade the metrics tested. The choice of radio duty cycle is very important to the network performance. The results show that the RPL performance is strictly dependent of the duty cycle.

One of the puzzling problems in mathematical biology is to show the existence of any form of error-correcting code in the DNA structure. Here we propose a model for the biological coding system similar to that of a digital communication system. This model consists of an encoder (a mapper and a BCH code over ℤ4) and a modulator (genetic code). Here we show that DNA sequences including proteins and targeting sequences from different species with 63, 255, and 1023 nucleotides long were identified as codewords of ℤ4-linear codes.

This paper proposes adaptive error control strategies for wireless sensor networks using informational value of messages. The informational value is based on sensors coverage area. Important packets are protected by more powerful error control schemes than less important packets. BCH codes and retransmission schemes were analyzed using OQPSK modulation in Rayleigh fading channels. The results obtained show that the adaptive schemes improve the reliability of packets with high informational value without a significant increase in the energy consumption.

This paper analyzes the energy efficiency of wireless sensor networks using different error control schemes. An analytical model is presented to evaluate the energy efficiency in Nakagami-ra fading channels. The model is applied to error control schemes of Bluetooth technology. Some custom error control schemes and adaptive techniques using different FEC and ARQ strategies are analyzed. Performance results are obtained through analysis for networks with different number of hops and fading channel conditions.

This paper studies different error control schemes in wireless sensor networks with Bluetooth technology. The tradeoff between reliability and energy consumption of Bluetooth packets are analyzed, using different error control techniques, such as retransmission and channel coding. The AUX1 packet is utilized for custom coding and also adaptive techniques are proposed based on the number of hops of the network. The wireless channel is modeled with Rayleigh fading. The results obtained may be used as references to determine the packet type in a sensor application

Bluetooth is a promising wireless technology designed for short-range ad hoc networks. This paper addresses issues related to Bluetooth performance in Nakagami-m fading channels. This distribution can model different fading intensities through parameter m. We evaluate the throughput and the packet loss rate of Bluetooth links using asynchronous packets. Relations between packet error rate and node distances are then derived, and a new piconet scheduling algorithm based on the estimated channel state information is then proposed. This polling scheme detects fading conditions and avoids transmission in bad channel conditions. The best strategy is defined using the received signal-to-noise ratio and the Nakagami fading parameter m. Simulation results demonstrate the algorithm performance for different traffic conditions.