WiFi sensing relies heavily on blackbox machine learning (ML) models due to the large feature space and complexity. Despite achieving very high accuracies in complex scenarios, the blackbox nature of these ML-based sensing techniques is commonly criticized. This is in fact a major source of mistrust as these models provide very little explanation supporting their decision, while often handling critical applications (e.g., elderly monitoring). In this paper, we investigate explainable artificial intelligence (XAI) techniques to shed light on the decisions and behaviors of such blackbox models. Specifically, we propose eXSense, a workflow designed based on state-of-the-art XAI techniques to analyze the behavior of blackbox models both locally and globally. To demonstrate its potential, we conduct an extensive analysis on two case studies from the recent sensing literature. Finally, leveraging the insights obtained from our analysis, we propose and evaluate changes to these models, thus enhancing their efficiency and reliability. This includes reducing the feature space by at least 80% with no/minimal loss ($\leq 1\%$) to the model accuracy.

This paper presents a novel approach to mitigate pilot contamination by enabling effective beamforming in crowded massive Multiple-Input-Multiple-Output (MIMO) systems operating in Time Division Duplex (TDD) mode. The main idea is to resolve angular overlaps among the signals that base stations receive from users in the wireless network. In this method, users randomly become active and try to access their corresponding base stations. Then, they periodically participate in a carefully designed game wherein only winners get to remain active. In this game, the probability of winning among a group of contenders is a direct function of having relatively higher transmission power, better channel conditions, and more correlated directions of arrival to the angles that the base stations estimate. We analytically prove that this game can guarantee the resolution of the so-called angular overlap, which enables effective beamforming and leads to pilot decontamination. We also support our statements through Monte Carlo simulations.

Mobility management is a major challenge for millimeter-wave (mmWave) cellular networks. In particular, directional beamforming in mmWave devices renders high-speed mobility support very complex. This complexity, however, is not limited to system design but also the performance estimation and evaluation. Hence, some have turned their attention to stochastic modeling of mmWave vehicular communication to derive closed-form expressions that can characterize the coverage and rate behavior of the network. In this article, we model and analyze the beam management for mmWave vehicular networks. To the best of our knowledge, this is the first work that goes beyond coverage and rate analysis. Specifically, we focus on a multi-lane divided highway scenario in which base stations and vehicles are present on both sides of the highway. In addition to providing analytical expressions for the average number of beam switching and handover events, we provide design insights for the operators to fine-tune their network through more informed choice of system parameters, including the number of resources dedicated to channel feedback and beam alignment operations.

Caching of the popular content in the storages of local nodes near to the users has been proposed as a promising approach to handle the network data traffic and the backhaul constraint of access points in the cellular networks. These local storages, also called caches or helpers nodes, are the nodes with the limited communication range but with considerable storage capacities which are generally established condensed to realise the above goals. This dense distribution makes possible the collaboration among the helper nodes for exchanging their embedded files and delivery of the requested files to the users. In this study, to manage the traffic in smaller areas and also to establish more users' accesses to their requested files, clustered dense helper scenario is considered. More precisely, the helper nodes are grouped in disjoint clusters, through the clustering strategy induced by Matern hard core point process. Then the closed‐form expressions based on stochastic geometry are derived for some performance metrics of the network. The analytical derivations are validated with Monte Carlo simulations which also compare two cases of clustered dense helper nodes and fewer helper nodes without clustering scenarios. The results show the advantage of the former from the outage probability perspective.

An overly multiple‐input–multiple‐output cognitive radio femtocell network consisting of P macrocell users equipments (MUEs), known as primary users (PUs), and K femtocell access points (FAPs), known as secondary users is considered in which femtocell user equipments (FUEs) operate in the closed access mode. Each FAP is equipped with a CR device to perform the local channel state information. The main objective is to maximise the average sum‐rate of the macrocell base station (MBS), as well as significantly increase in the FUE’s rate with negligible reduction in the PU’s rate for moderate‐to‐high signal‐to‐noise ratio regimes. This goal is achieved by proposing an opportunistic interference alignment (OIA) technique using the threshold‐based beamforming (TBF) algorithm. Assuming that the receiver and the transmitter of the PU have perfect knowledge of their own channel matrices, the authors use the maximum eigenmode beamforming algorithm for the transmission between the MBS and its MUEs, where the PU releases some of its eigenmodes for the FAPs, in order to transmit signals for their corresponding FUEs over them. To decide on the presence or absence of MUEs, they use the generalised likelihood ratio test detector which is more robust to the noise uncertainty than the energy detector. The proposed OIA‐TBF protocol allows the opportunistic FAPs to send data for FUEs and use the same frequency band of a preexisting MUE to guarantee that no interference is imposed on the MUE’s performance for such a network.

Employing femtocell technology noticeably improves both the indoor coverage and the overall cellular network's capacity. In the downlink scenario, when macrocell users (MUs) are far from the macro base station (MBS) and close to the femto access points (FAPs), the transmitted signals of FAPs to their users cause severe interference on MUs. One solution to overcome this issue is serving MUs by neighbouring FAPs. Generally femtocells are selfish and have no desire to allocate their resources to MUs. In this study, a framework to motivate femtocells to share their free resources with MUs is proposed. In this framework, the femtocells are assumed to belong to a specific unit. The unit encourages and manages the femtocells to share their resources, and in exchange receives a monetary utility from the MBS, and as a result improves the performance of the network. A computationally‐efficient algorithm is proposed, which approximates the complete search algorithm, and their solutions coincide with each other in many cases. After FAPs assignment, the authors try to allocate the optimal power to MUs served by each femtocell to further improve their coverage probability through a bargaining game. The simulation results verify the performance of the proposed model and confirm the coverage improvement obtained in the network.

Multi-tier cellular communication networks constitute a promising approach to expand the coverage of cellular networks and enable them to offer higher data rates. In this paper, an uplink two-tier communication network is studied, where macro users, femto users and femto access points are geometrically located inside the coverage area of a macro base station according to Poisson point processes. Each femtocell is assumed to have a fixed backhaul constraint that puts a limit on the maximum number of femto and macro users it can service. Under this backhaul constraint, the network adopts a special open access policy, in which each macro user is either assigned to its closest femto access point or to the macro base station, depending on the ratio between its distances from those two. Under this model, upper and lower bounds on the outage probabilities experienced by users (macro and femto) serviced by femto access points are derived as functions of the distance between the macro base station and the femto access point serving them. The bounds are confirmed via simulation results.

With growing demand for high data rates, the mobile network has faced excessive traffic and it seems that one of the few long‐term solutions to this problem is employing femtocells which improve capacity and indoors coverage. In the typical femtocell networks, co‐ and cross‐layer interferences caused by simultaneous transmission of macrocell base stations and femtocell access points in the same frequency band, results in dead spots especially if femtocells are densely employed. In this study, to mitigate the interference and improve the system's performance, the authors study the application of time‐hopping approach. Particularly, the impact of aggressive signals of femto access points on a macro user equipment located in the second tier, that is, the femtocell coverage area, is analysed and the semi‐closed‐form expression for the system outage probability is derived. Based on the analysis, they also simulate the transmission capacity (TC) that represents the total capacity of the co‐channel two‐tier networks with outage constraints. Next, employing Gaussian approximations, they derive upper and lower bounds on the outage probability. These bounds provide better insight on the effect of system's parameters on its performance. Their numerical results show that the time‐hopping technique significantly reduces the outage probability and therefore increases the TC of two‐tier networks.

Employing multi-tier networks is among the most promising approaches to address the rapid growth of the data demand in cellular networks. In this paper, we study a two-tier uplink cellular network consisting of femtocells and a macrocell. Femto base stations, and femto and macro users are spatially deployed based on independent Poisson point processes. We consider an open access assignment policy, where each macro user based on the ratio between its distances from its nearest femto access point (FAP) and from the macro base station (MBS) is assigned to either of them. By tuning the threshold, this policy allows controlling the coverage areas of FAPs. For a fixed threshold, femtocells coverage areas depend on their distances from the MBS; those closest to the fringes will have the largest coverage areas. Under this open-access policy, ignoring the additive noise, we derive analytical upper and lower bounds on the outage probabilities of femto users and macro users that are subject to fading and path loss. We also study the effect of the distance from the MBS on the outage probability experienced by the users of a femtocell. In all cases, our simulation results comply with our analytical bounds.

In this paper, we present the bit error probability (BEP) of a relay network employing UWB links between its nodes. Since UWB systems can resolve many paths and are thus rich in multipath diversity, the use of the Rake diversity combining is very effective. However, because of the complexity, it is impractical in the scenarios with portable terminals or in the networks with multiple relays. Time-reversal technique, in which the transmitting nodes can do the pre-diversity filtering of UWB signal before transmission, is known as an approach to achieve the performance equivalent to the Rake combining without increasing the receiver complexity. The performance of such a network with simple correlators for each link at the receiver side is analyzed considering some realistic channel estimation errors. The results show that there is a great potential in combining time reversal and UWB techniques with respect to both reducing the receiver complexity and improving the system performance. The system robustness to imperfect channel estimation is also shown in the numerical analysis and compared to traditional UWB system with All-rake receiver. The analytical expressions are also validated by computer simulations which confirm the accuracy of the approximations used in obtaining these results.

In this paper, the extension of cooperative communication to the context of TH-UWB is investigated. In particular, the average bit error probability (BEP) is provided for cooperative TH-UWB systems with decode-and-forward relaying protocol. In the considered relay network, UWB links among the nodes are modeled according to IEEE 802.15.4a standards. Our methodology is based on computing the characteristic function (CF) of the decision variable at the destination terminal. We use Gaussian quadrature numerical method to approximate the CF of interference component appeared in decision variable term. This technique permits to predict the system performance in different IEEE 802.15.4a defined channels with high accuracy and reasonable complexity. Numerical results show that significant improvement in the BEP of impulse radio UWB system is obtained by adding a relay node. The analytical expressions are also validated by computer simulations which confirm the accuracy of the approximations used in obtaining the BEP.

We derive bounds on the expected capacity and outage capacity of a three-node relay network for UWB communications. We also provide a simple tight approximation for the derived upper bound on the capacity and then using this bound we obtain the outage probability of the network. Numerical results show that a significant improvement in the system capacity and outage probability is obtained by adding a relay node. Moreover, our theoretical results reveal that the diversity gain of a relay channel substantially increases by using UWB links instead of NB links. We also derive these bounds when we have a constraint on the total transmitted power of the source and the relay nodes.

A cooperative space–frequency (SF) coded orthogonal frequency division multiplexing system is considered and its performance over quasi-static frequency-selective Rayleigh fading channels is evaluated. An expression for exact outage error probability is derived and its tight closed-form lower bound is presented. The tightness of the lower bound is demonstrated through Monte Carlo simulation. Asymptotic analysis indicates that the proposed protocol achieves full spatial and frequency diversities available in the cooperative communication system. The theoretical analysis of the proposed SF coded cooperation protocol is further confirmed by computer simulation using a previously introduced SF block code that is capable of achieving full spatial and frequency diversities.

Traditionally, the frequency spectrum is licensed to users by government agencies in a fixed manner where the licensee has exclusive right to access the allocated band. However, with increasing demand for the spectrum and scarcity of vacant bands, a spectrum policy reform seems inevitable. Meanwhile, recent measurements suggest the possibility of sharing spectrum among different parties subject to interference-protection constraints. In order to enable access to an unused licensed spectrum, a secondary user has to monitor licensed bands and opportunistically transmit whenever no primary signal is detected. Spectrum-sharing between a primary licensee and a group of secondary users has been studied. The structure of an asymptotically optimum detector based on the measurements of all secondary users is derived and the effect of the quantisation error in such a system is evaluated. Also, it is shown that by using the proposed detector in a sequential detection structure, it is possible to shorten the decision time needed by the detector. The results show the superiority of the proposed detector to other schemes.

We evaluate and compare the multiuser performance of the multicarrier frequency-hopping (MC-FH) CDMA system using coherent and non-coherent detection in AWGN and slowly frequency-selective Rayleigh fading channels in a downlink application. For fair comparison, we use BFSK modulation for both cases. Our numerical results indicate that in both AWGN and fading channels, coherent detection substantially outperforms non-coherent detection. Furthermore, it has been realized that the performance enhancement due to the diversity order is only observable for coherent detection.