Due to the rapidly evolving landscape of cybersecurity, the risks in securing cloud networks and devices are attesting to be an increasingly prevalent research challenge. Reinforcement learning (RL) is a subfield of machine learning that has demonstrated its ability to detect cyberattacks, as well as its potential to recognize new ones. Many of the popular RL algorithms at present rely on deep neural networks, which are computationally very expensive to train. An alternative approach to this class of algorithms is hyperdimensional computing (HDC), which is a robust, computationally efficient learning paradigm that is ideal for powering resource-constrained devices. In this article, we present CyberRL, a HDC algorithm for learning cybersecurity strategies for intrusion detection in an abstract Markov game environment. We demonstrate that CyberRL is advantageous compared to its deep learning equivalent in computational efficiency, reaching up to$1.9{\times }$speedup in training time for multiple devices, including low-powered devices. We also present its enhanced learning quality and superior defense and attack security strategies with up to$12.8\times $improvement. We implement our framework on Xilinx Alveo U50 FPGA and achieve approximately$700\times $speedup and energy efficiency improvements compared to the CPU execution.

In an era where many digital interactions are often short-lived and instantaneous, connectionless communication technologies have been emerging as pivotal enablers of such interactions. Unlike their connection-oriented counterparts, these technologies facilitate seamless, on-demand exchanges without the need for establishing persistent connections. Relying on technologies that are abundant and easily accessible for the end users contributes to the ubiquitous spread of connectionless communication. WiFi and BLE are two popular examples of such technologies as they are found now on all user devices. This paper explores their potential in realizing connectionless communication scenarios by experimentally comparing their usages for several use-cases in different application domains. Such a comparison is imperative to making an informed decision on which technology to use for a given application. We establish this comparison on ESP32 in terms of power consumption, delivery rate and distance.

Breakthroughs in deep learning (DL) continuously fuel innovations that profoundly improve our daily life. However, DNNs overwhelm conventional computing architectures by their massive data movements between processing and memory units. As a result, novel computer architectures are indispensable to improve or even replace the decades-old von Neumann architecture. Nevertheless, going far beyond the existing von Neumann principles comes with profound reliability challenges for the performed computations. This is due to analog computing together with emerging beyond-CMOS technologies being inherently noisy and inevitably leading to unreliable computing. Hence, novel robust algorithms become a key to go beyond the boundaries of the von Neumann era. Hyper-dimensional Computing (HDC) is rapidly emerging as an attractive alternative to traditional DL and ML algorithms. Unlike conventional DL and ML algorithms, HDC is inherently robust against errors along a much more efficient hardware implementation. In addition to these advantages at hardware level, HDC's promise to learn from little data and the underlying algebra enable new possibilities at the application level. In this work, the robustness of HDC algorithms against errors and beyond von Neumann architectures are discussed. Further, the benefits of HDC as a machine learning algorithm are demonstrated with the example of outlier detection and reinforcement learning.

Cybersecurity has emerged as a critical challenge for the industry. With the large complexity of the security landscape, sophisticated and costly deep learning models often fail to provide timely detection of cyber threats on edge devices. Brain-inspired hyperdimensional computing (HDC) has been introduced as a promising solution to address this issue. However, existing HDC approaches use static encoders and require very high dimensionality and hundreds of training iterations to achieve reasonable accuracy. This results in a serious loss of learning efficiency and causes huge latency for detecting attacks. In this paper, we propose CyberHD, an innovative HDC learning framework that identifies and regenerates insignificant dimensions to capture complicated patterns of cyber threats with remarkably lower dimensionality. Additionally, the holographic distribution of patterns in high dimensional space provides CyberHD with notably high robustness against hardware errors.

Reinforcement Learning (RL) has opened up new opportunities to enhance existing smart systems that generally include a complex decision-making process. However, modern RL algorithms, e.g., Deep Q-Networks (DQN), are based on deep neural networks, resulting in high computational costs. In this paper, we propose QHD, an off-policy value-based Hyperdimensional Reinforcement Learning, that mimics brain properties toward robust and real-time learning. QHD relies on a lightweight brain-inspired model to learn an optimal policy in an unknown environment. On both desktop and power-limited embedded platforms, QHD achieves significantly better overall efficiency than DQN while providing higher or comparable rewards. QHD is also suitable for highly-efficient reinforcement learning with great potential for online and real-time learning. Our solution supports a small experience replay batch size that provides 12.3 times speedup compared to DQN while ensuring minimal quality loss. Our evaluation shows QHD capability for real-time learning, providing 34.6 times speedup and significantly better quality of learning than DQN.

Traditional robot control or more general continuous control tasks often rely on carefully hand-crafted classic control methods. These models often lack the self-learning adaptability and intelligence to achieve human-level control. On the other hand, recent advancements in Reinforcement Learning (RL) present algorithms that have the capability of human-like learning. The integration of Deep Neural Networks (DNN) and RL thereby enables autonomous learning in robot control tasks. However, DNN-based RL brings both high-quality learning and high computation cost, which is no longer ideal for currently fast-growing edge computing scenarios.

Reinforcement Learning (RL) is a powerful technology to solve decisionmaking problems such as robotics control. Modern RL algorithms, i.e., Deep Q-Learning, are based on costly and resource hungry deep neural networks. This motivates us to deploy alternative models for powering RL agents on edge devices. Recently, brain-inspired Hyper-Dimensional Computing (HDC) has been introduced as a promising solution for lightweight and efficient machine learning, particularly for classification.

In this paper, we present Hyperdimensional Hybrid Learning (HDHL), which combines model-free and model-based Reinforcement Learning, to effectively reduce the computational cost and environment interaction for optimizing an intelligent cloud service. We first show that Hyperdimensional Q-Learning (QHD), the state-of-the-art Hyperdimensional Computing value-based Reinforcement Learning algorithm, is computationally faster than the Deep Q-Network (DQN) for this task. In addition, we demonstrate how HDHL reduces the number of environment interactions by 4.8× to learn the near optimal configuration. Our evaluation shows that HDHL is computationally more efficient than both Q-Learning algorithms, with the total time being reduced by 21.0× compared to DQN and 16.5× compared to QHD.

In the recent year, people had to work from home due to the outbreak of the COVID-19 pandemic. When the majority of the family members are working online, the bitrate experienced by the average user may drop, especially if some members have to work in rooms that suffer from weak coverage. Benefiting from the emerging concept of Reflective Intelligent Surfaces (RIS), the network coverage in our houses can be greatly improved. This paper presents a study of an RIS-assisted system for an indoor scenario operating at 2.4GHz. We propose an RIS placement approach that is based on minimizing the pathloss of the channel, to enhance the rate of bad coverage rooms, while taking into consideration their user occupancies. The proposed approach, which we refer to as the Weighted RIS Placement, is modeled and simulated for a single RIS. The problem is then extended to a two-RIS scenario. Our results show that the Weighted RIS placement provides significant rate gains. Also, this is the first work that models the communication channels for the individual rooms, using corresponding Rician K-factor values that reflect the indoor layout.

Carrier Aggregation (CA) has been evolving with all releases of LTE due to its large benefits. However, LTE is based on OFDM which suffers from its high Peak-to-Average-Power-Ratio (PAPR), therefore inducing non-linearities and causing deterioration in the power efficiency of the High Power Amplifier (HPA). Although the PAPR problem has been extensively tackled, there remains a need for techniques that can cope with current system designs. We therefore propose in this paper a new distortion-less PAPR reduction method that is characterized by its high performance and significantly lower complexity when compared to existing methods. We refer to this technique as the Peak Valley Alignment (PVA) method. It is inspired from the sequence alignment techniques used in bioinformatics. We show that the performance attained by optimization-based methods can be approximately achieved through the use of heuristics at a much lower complexity.

Carrier Aggregation started to appear since release 10 of LTE, and offers higher data rates, but at the cost of increased Peak-to-Average-Power-Ratio (PAPR). This paper studies the distribution of PAPR through analyzing the Crest Factor (CF) which is the square root of the PAPR of the Carrier Aggregated OFDM (CA-OFDM) signals used in the downlink of LTE-A. Although the base station is not power constrained, the increased PAPR can adversely affect performance in terms of energy efficiency and coverage. We model the distribution of the CF by adopting the Level Crossing Rate analysis to approximate the peak distribution of the CA-OFDM signal which can be characterized as a bandlimited complex Gaussian process. The expression of the distribution is derived for an arbitrary number of component carriers for both contiguous and non-contiguous aggregation, where simulations show good agreement with the derived analytical expressions. The derived expression is applied to model the Selective Mapping technique used for PAPR reduction.