Walaa Hamouda’s research while affiliated with Concordia University and other places

The next frontier in wireless connectivity lies at the intersection of cognitive radio (CR) technology and machine learning (ML), where intelligent networks can provide pervasive connectivity for an ever-expanding range of applications. In this regard, this survey provides an in-depth examination of the integration of ML-based CR in a wide range of emerging wireless networks, including the Internet of Things (IoT), mobile communications (vehicular and railway), and unmanned aerial vehicle (UAV) communications. By combining ML-based CR and emerging wireless networks, we can create intelligent, efficient, and ubiquitous wireless communication systems that satisfy spectrum-hungry applications and services of next-generation networks. For each type of wireless network, we highlight the key motivation for using intelligent CR and present a full review of the existing state-of-the-art ML approaches that address pressing challenges, including energy efficiency, interference, throughput, latency, and security. Our goal is to provide researchers and newcomers with a clear understanding of the motivation and methodology behind applying intelligent CR to emerging wireless networks. Moreover, problems and prospective research avenues are outlined, and a future roadmap is offered that explores possibilities for overcoming challenges through trending concepts.

Modulation recognition is one of the key elements in the cognitive radio (CR) technology. Chiefly, automatic modulation recognition (AMR) is a challenging task in such systems. It refers to blindly identifying the modulation type by a CR receiver which is highly recommended for military as well as civil applications, e.g. security, threat analysis. In this paper, a comprehensive testbed is provided. It aims at assessing the AMR performance in near real-world multiple-input multiple-output (MIMO) communication scenario, as a first step for industrial integration and commercialization. Our testbed is based on the Software-Defined Radio (SDR) technology where the radio reconfigurability feature is warranted by integrating the NI-2954R Universal Software Radio Peripheral (USRP) reconfigurable I/O (RIO) boards. The different processing modules of the designed testbed have been implemented within a scalable and flexible architecture to realize real-time recognition. It also serves as an efficient tool to bridge the gap between theory and practice. Empirical results showed that the proposed testbed yields high recognition accuracy and strong robustness.

Lossless turbo source coding with decremental/incremental redundancy is a variable-length source coding scheme which employs turbo codes for data compression. Although the scheme offers low compression rates and lends itself to joint source-channel coding, it suffers from a large delay in the encoding phase. The delay is imposed by several tentative encoding-decoding procedures performed at the encoder to search for the minimum compression length. In this work, we apply machine learning to provide a highly accurate estimate of the proper compression length. The encoder starts its search from this estimated length, thus the delay of turbo source coding will decrease considerably. The preliminary results show a four-fold reduction in the encoding delay at the expense of a negligible increase in the compression rate.

In this paper, the physical-layer security (PLS for an underlay cognitive radio network (CRN with multiple non colluding eavesdroppers is studied. We assume that a secondary user (SU transmitter communicates with an SU receiver over a cascaded -fading channel and under the threat of eavesdropping. A cooperative jammer exists to improve the SUs security by harvesting energy from the SU transmitter using power splitting technique. This harvested energy is utilized to generate jamming signals to deteriorate the eavesdroppers reception quality. Moreover, the eavesdroppers are assumed to be randomly distributed according to a homogeneous Poisson point process (HPPP, in which the kth closest eavesdropper to the SU transmitter will be selected in our analysis. Additionally, the legitimate receiver is assumed to harvest energy from the SU transmitter using power splitting technique. In this context, we propose an optimum power splitting factor at the legitimate receiver that achieves the best privacy while constraining the amount of the harvested energy. Furthermore, on enhancing system security, we investigate the effectiveness of the jamming technique. The impact of distances over security is also examined. The results reveal an improvement in privacy as the legitimate receiver utilizes the adaptable power splitting factor. PLS is evaluated in terms of the probability of non-zero secrecy capacity and the intercept probability.

Heterogeneous cloud radio access network (H-CRAN) has been deemed as a promising architecture for the fifth generation wireless networks. In H-CRAN, small remote radio heads (SRRHs) are deployed underlaying the macro remote radio head (MRRH) to boost the spectral efficiency. H-CRANs with dense RRHs suffer from severe inter-cell interference due to the random deployment of dense SRRHs, which limits its performance. In this paper, we propose an Edge-Aware RRH-Cooperation (EARC) scheme to mitigate the impact of both the cross-tier and co-tier interference. We partition the users into two groups, Cell Edge Device (CED) and Non-cell Edge Device (NED). These two groups, CED and NED, are mapped into two association modes, dual-association and single-association mode, respectively. In single-association mode, the NED associates with the RRH that provides the maximum received signal. On the other hand, in the dual-association mode, the CED associates with the two strongest RRHs where the two RRHs may belong to the same or different tier(s). Using tools from stochastic geometry, we quantify the performance gains of the proposed EARC scheme in terms of outage probability and ergodic rate. To further show the efficacy of the proposed scheme, we compare it to four schemes developed in the literature.

In this paper, the physical-layer security for a three-node wiretap system model is studied. Under the threat of multiple eavesdroppers, it is presumed that a transmitter is communicating with a legitimate receiver. The channels are assumed to be following cascaded κ-μ fading distributions. In addition, two scenarios for eavesdroppers’ interception and information-processing capabilities are investigated: colluding and non-colluding eavesdroppers. The positions of these eavesdroppers are assumed to be random in the non-colluding eavesdropping scenario, based on a homogeneous Poisson point process (HPPP). The security is examined in terms of the secrecy outage probability, the probability of non-zero secrecy capacity, and the intercept probability. The exact and asymptotic expressions for the secrecy outage probability and the probability of non-zero secrecy capacity are derived. The results demonstrate the effect of the cascade level on security. Additionally, the results indicate that as the number of eavesdroppers rises, the privacy of signals exchanged between legitimate ends deteriorates. Furthermore, in this paper, regarding the capabilities of tapping and processing the information, we provide a comparison between colluding and non-colluding eavesdropping.

The cloud radio access network (CRAN) is a promising architecture that can face the hard challenges imposed on the next cellular generations. However, the high computational complexity of the optimization techniques currently proposed in the literature prevents from totally reaping the benefits of the CRAN architecture. Learning based techniques are expected to replace the conventional optimization techniques due to their high performance and very low online computational complexity. In this paper, we consider tackling the power allocation in CRAN via an unsupervised deep learning based approach. Different from the previous works, user association is considered in our optimization problem to reflect a real cellular scenario. Additionally, we propose a novel scheme that can enhance the deep learning based power allocation approaches, significantly. We provide intensive analysis to discuss the trade-offs faced when employing our deep learning based approach for power allocation. Simulation results prove that the proposed technique can obtain a very close to optimal performance with negligible computational complexity.

This paper investigates the physical-layer security (PLS) of a single-input-multiple-output (SIMO) underlay cognitive radio network (CRN) over cascaded $\kappa$ - $\mu$ fading channels. A secondary user (SU) transmitter communicates with a secondary user receiver over cascaded $\kappa$ - $\mu$ channels. Moreover, an eavesdropper residing in the SUs transmission’s range would be capable of intercepting the SUs’ transmission. The SU destination and the eavesdropper are both assumed to be equipped with multiple antennas. Both of the receivers adopt maximal-ratio combining (MRC) over the multiple copies of the signal. In an underlay CRN, the SU transmitter should keep adjusting the transmission power to ensure not to disturb the primary users’ (PUs) transmission. PLS is analyzed in terms of the secrecy outage probability ( SOP ) and the probability of non-zero secrecy capacity $(P_{nz})$ . Results indicate the evident effect of the cascade level and the number of antennas at the eavesdropper over the secrecy of the SUs pair. In addition, results reveal that PLS can be strengthened by increasing the number of antennas at the legitimate receiver. The impact of the interference level tolerable by the PU receiver over SUs’ secrecy is also analyzed. Monte-Carlo simulations and analytical results are presented to assess the system performance.

Cell-free (CF) massive multiple-input multiple-output (MIMO) system is currently considered as a promising network architecture to satisfy the anticipated rate requirements of beyond-5G networks. However, in practical scenarios with the presence of high-velocity users, the network experiences an inevitable performance degradation due to the Doppler shift effect. This paper analyzes the potential of frame length optimization in limiting the Doppler shift effect on the performance of time-division duplexing CF massive MIMO under different mobility conditions. In doing so, we derive novel expressions for tight lower bound of the average downlink (DL) and uplink (UL) rates. Capitalizing on the derived analytical results, we provide an analytical framework to determine the optimal frame length that limits the Doppler shift effect on DL and UL rates according to some criterion. Our results show perfect match of both analytical and simulated results under different system settings. Also, we reveal that the optimal frame lengths for maximizing the DL and UL rates are different and depend mainly on the transmission criterion and the users’ velocities. Besides, our results demonstrate the high potential of adapting the frame length according to the velocity conditions in limiting the Doppler shift effect, compared to applying a fixed frame length.

This paper studies the physical-layer security (PLS)of a system model consisting of a transmitter, a receiver, and multiple eavesdroppers. Cascaded general fading channel, which is the κ-μdistribution is assumed at the main and the wiretap links of the network. The impacts of the cascade level, the number of eavesdroppers attempting to overhear the confidential information, and the wiretap channel’s parameters on the system’s secrecy are investigated. Two of the main secrecy metrics are used to evaluate the secrecy level of the system, which are the secrecy outage probability (OPsec)and the probability of non-zero secrecy capacity (Pnsc). Exact and asymptotic form expressions for (OPsec) and (Pnsc)are derived. Asymptotic analysis is performed to gain a clear vision about the impact of some key parameters over the secrecy. The results show that the fading channel cascade level has a significant effect on the system’s secrecy. Also, the results show that the system is less protected when increasing the number of eavesdroppers or when improving the wiretap channel’s conditions. Analytical results are validated using Monte-Carlo simulations.