Abstract Intelligent reflective surface (IRS) provides an effective solution for reconfiguring air‐to‐ground wireless channels, and intelligent agents based on reinforcement learning can dynamically adjust the reflection coefficient of IRS to adapt to changing channels. However, most exiting IRS configuration schemes based on reinforcement learning require long training time and are difficult to be industrially deployed. This paper, proposes a model‐free IRS control scheme based on reinforcement learning and adopts transfer learning to accelerate the training process. A knowledge base of the source tasks has been constructed for transfer learning, allowing accumulation of experience from different source tasks. To mitigate potential negative effects of transfer learning, quantitative analysis of task similarity through unmanned aerial vehicle (UAV) flight path is conducted. After identifying the most similar source task to the target task, parameters of the source task model are used as the initial values for the target task model to accelerate the convergence process of reinforcement learning. Simulation results demonstrate that the proposed method can increase the convergence speed of the traditional DDQN algorithm by up to 60%.

Convolutional neural networks (CNNs) are used comprehensively in the field of the Internet of Things (IoTs), such as mobile phones, surveillance, and satellite. However, the deployment of CNNs is difficult because the structure of hand-designed networks is complicated. Therefore, we propose a sensitiveness-based network pruning framework (SNPF) to reduce the size of original networks to save computation resources. SNPF will evaluate the importance of each convolutional layer by the reconstruction of inference accuracy when we add extra noise to the original model, and then remove filters in terms of the degree of sensitiveness for each layer. Compared with previous weight-norm-based pruning methods, such as “$\mathscr {C}_{1}$-norm,” “BatchNorm-Pruning,” and “Taylor-Pruning,” SNPF is robust to the update of parameters, which can avoid the inconsistency of evaluation for filters if the parameters of the pretrained model are not fully optimized. Namely, SNPF, can prune the network at the early training stage to save computation resources. We test our method on three prevalent models of VGG-16, ResNet-18, ResNet-50 and a customized Conv-4 with 4 convolutional layers. They are then tested on CIFAR-10, CIFAR-100, ImageNet, and MNIST, respectively. Impressively, we observe that even when the VGG-16 is only trained with 50 epochs, we can get the same evaluation of layer importance as the results when the model is fully trained. Additionally, we can also achieve comparable pruning results to previous weight-oriented methods on the other three models.

Unmanned aerial vehicle (UAV) carrying intelligent reflecting surface (IRS) can serve as an aerial platform to improve the coverage area and transmission performance of traditional wireless network. In this paper, we investigate the weighted sum rate maximization problem for a UAV-assisted and IRS-based multi-user communication system. Specifically, by applying the alternating optimization approach, we propose a two-phase approach to effectively optimize the number of activated reflective elements, the precoding matrix, the phase shift, and the UAV location. Numerical results validate that the proposed approach can converge at a faster rate and improve the weighted sum rate performance.

In this paper, we investigate the secrecy outage probability (SOP) for a multi-tier unmanned aerial vehicular (UAV) assisted cognitive communication network. Specifically, the low-altitude rotary-wing (LARW) UAV relays harvest radio frequency (RF) energy from the transmission of a high-altitude fixed-wing (HAFW) UAV and forward the information to a ground destination under a decode-and-forward protocol while a ground eavesdropper tries to capture the relay signal. The multi-tier UAV system transmits in an underlay fashion over a licensed spectrum of a primary user. We study the exact statistics of SOP assuming that the number of UAV relays follows a Poisson point process (PPP). The simulation results are presented to illustrate and verify the analytical results.

In this work, we study the performance of reconfigurable intelligent surface (RIS)-aided communication systems over the sum of cascaded Rician fading channels. To facilitate the performance analysis of the practical scenario, we consider the case that imperfect channel state information (CSI) is available at the RIS. We derive the closed-form expressions of several performance metrics in terms of the exact outage probability, ergodic capacity and average bit error rate (BER). Through analytical and numerical results, we examine the effect of the number of reflecting elements at the RIS and different system parameters on the overall system performance.

Due to concurrent access attempts from massive machine-type devices (MTDs) within the wide beam coverage, the existing contention-based random access (RA) scheme suffers from severe physical random access channel (PRACH) over-load when applied to the emerging satellite-enabled machine-to-machine (M2M) communications. In this paper, we propose an efficient collision detection scheme based on cyclically shifted Zadoff-Chu (ZC) sequences, which are generated by the minimum number of required root indexes and a fixed cyclic shift offset independent of the beam radius. The proposed scheme enables rapid collision detection at the first step of RA procedure by capturing correlation peaks at the timing positions corresponding to the multiples of the cyclic shift offset, thus can reduce the access delay and resource consumptions for the collided MTDs. Simulations are carried out to validate the correctness of mathematical analysis, and to demonstrate the significant detection performance improvement of our scheme with effective non-orthogonal interference (NOI) mitigation by compared to the conventional one.

An energy-efficient two-way (TW) full-duplex (FD) network with the assistance of an unmanned aerial vehicle (UAV) is proposed, where the UAV acts as a mobile relay to assist the information exchange between two terrestrial transceivers. In particular, the self-interference (SI) channel gains follow complex Gaussian distribution and the perfect channel state information (CSI) of SI channels is unavailable at the receiver. To maximize the energy efficiency (EE), UAV flight speeds are controlled and power adaptation at the UAV relay is performed. The genetic algorithm (GA) is applied to efficiently obtain the optimal solution. Numerical results show that our scheme performs better than the one-way (OW) FDR scheme, fixed power (FP) and fixed flight speed (FS) policy. In addition, the SI cancellation factor on the EE is also demonstrated.

With adaptive transmission, the transmission rate and/or power are adaptively adjusted according to the channel realization, which leads to a variable amount of energy consumption for the transmission of the same amount of data. We propose an analytical framework to statistically characterize the transmitter energy consumption of adaptive transmission over fading channels. For slow fading scenario, we derive the probability density function (PDF) and cumulative distribution function (CDF) of energy consumption assuming continuous-time Markov channel model. For the fast fading case, we apply the statistical mixture model to obtain the approximate PDF of energy consumption. Selected numerical results are presented to illustrate and to validate the mathematical formulations.

Radio frequency (RF) energy harvesting provides wireless sensors with permanent and convenient energy supply. In this paper, we study the statistics of harvested RF energy of secondary wireless sensor over quasi-static fading channels. We also investigate the performance of secondary sensor transmission with harvested RF energy. Specifically, assuming that the sensor uses all the harvested RF energy for transmission, we derive the exact statistics of received signal-to-noise ratio (SNR) over Nakagami fading channel. We also investigate the statistics of received SNR under a primary interference constraint. The statistics are applied to performance evaluation in term of outage probability and average error rate. Selected numerical results are presented to illustrate the mathematical formulation.

In cognitive radio communication system, unlicensed secondary user (SU) can opportunistically transmit over underutilized primary user's (PU) spectrum. With interweave implementation, SU performs spectrum sensing on the target frequency band to explore transmission opportunity. Sensing errors can greatly affect the performance of secondary transmission. In this paper, we propose a seven-state discrete-time Markov model to characterize slotted secondary transmission process with imperfect spectrum sensing. A two-dimensional Markov model is also developed to study the queueing performance of slotted secondary transmission with adaptive modulation and coding. We show that false alarm has significant effect on the secondary throughput, whereas miss detection only notably reduces the secondary throughput when the traffic intensity is low.

In this paper, we investigate the transmission time of a large amount of data over fading wireless channel with adaptive modulation and coding (AMC). Unlike traditional transmission systems, where the transmission time of a fixed amount of data is typically regarded as a constant, the transmission time with AMC becomes a random variable, as the transmission rate varies with the fading channel condition. To facilitate the design and optimization of wireless transmission schemes for big data applications, we present an analytical framework to determine statistical characterizations for the transmission time of big data with AMC. In particular, we derive the exact statistics of transmission time over block fading channels. The probability mass function and the cumulative distribution function of transmission time are obtained for both slow and fast fading scenarios. We further extend our analysis to the Markov channel, where transmission time becomes the sum of a sequence of exponentially distributed time slots. Analytical expression for the probability density function of transmission time is derived for both fast fading and slow fading scenarios. These analytical results are essential to the optimal design and the performance analysis of future wireless transmission systems for big data applications.

Cognitive radio communications can opportunistically access underutilized spectrum for emerging wireless applications. With interweave cognitive implementation, secondary user transmits only if primary user does not occupy the channel and waits for transmission otherwise. Therefore, secondary packet transmission involves both transmission time and waiting time. The resulting extended delivery time (EDT) is critical to the throughput analysis of secondary system. In this paper, we study the EDT of secondary packet transmission with adaptive modulation under interweave implementation to facilitate the delay and throughput analysis of such cognitive radio system. In particular, we propose an analytical framework to derive the probability density functions of EDT considering random-length transmission and waiting slots. We also present selected numerical results to illustrate the mathematical formulations and to verify our analytical approach.

In this paper, we investigate the statistics of packet transmission time of wireless transmission systems employing adaptive modulation. Unlike traditional transmission systems, where the transmission time of a fixed-size packet is typically regarded as a constant, the transmission time of adaptive modulation systems depends on the channel realization as the transmission rate varies with the fading channel conditions. In this paper, we derive the exact statistical distribution of packet transmission time for adaptive modulation systems over block fading channels. The exact expressions of the probability mass function (PMF) and cumulative distribution function (CDF) of packet transmission time are obtained for both slow and fast fading scenarios. We further present an approximate PMF for fast fading scenario to reduce the computation complexity. Selected numerical results are presented to illustrate the mathematical formulation.