Hybrid beamforming for extremely large-scale multiple-input multiple-output (XL-MIMO) systems is challenging in the near field because the channel depends jointly on angle and distance, and the multiuser interference (MUI) is strong. Existing deep learning methods typically follow either a decoupled design that optimizes analog beamforming without explicitly accounting for MUI, or an end-to-end (E2E) joint analog–digital optimization that can be unstable under nonconvex constant-modulus (CM), pronounced analog–digital coupling, and gradient pattern of sum-rate loss. To address both issues, we develop a complex-valued E2E framework based on a variant minimum mean square error (variant-MMSE) criterion, where the digital precoder is eliminated in closed form via Karush–Kuhn–Tucker (KKT) conditions so that analog learning is trained with a stable objective. The network employs a grouped complex-convolution sensing front-end for uplink (UL) measurements, a shared complex multi-layer perceptron (MLP) for per-user feature extraction, and a merged constant-modulus head to output the analog precoder. In the indirect mode, the network designs hybrid beamformers from estimated channel state information (CSI). In the direct mode where explicit CSI is unavailable, the network learns the sensing operator and the analog mapping from short pilots, after which additional pilots estimate the equivalent channel and enable a KKT closed-form digital precoder. Simulations show that the indirect mode approaches the performance of iterative variant-MMSE optimization with a complexity reduction proportional to the antenna number. In the direct mode, the proposed method improves spectral efficiency over sparse-recovery pipelines and recent deep learning baselines under the same pilot budget.

In heterogeneous computing systems, excellent task scheduling algorithms can shorten the task completion time and improve system parallelism. With the large-scale deployment of edge computing, the task scheduling algorithm in heterogeneous edge computing servers has become a critical factor in improving the overall system performance. This paper proposes a new task scheduling algorithm called the edge cover scheduling algorithm (ECSA), which schedules tasks based on the edge cover queue of the directed acyclic graph (DAG) for heterogeneous computing systems. Based on the estimation of distribution algorithm (EDA) and the graph random walk algorithm, the ECSA generates an edge cover queue from DAG. Then, the ECSA uses the heuristics greedy method with low time and computational complexity to allocate the edge cover queue to processors. Theoretical analysis and simulation results on random DAGs and real-world DAGs show that the ECSA can achieve better scheduling results in terms of makespan, the schedule length ratio (SLR), efficiency, and frequency of best results with low time and computational complexity.

Due to the open nature of wireless communication, malicious electromagnetic jamming has long been a severe threat to the establishment and stability of communication links. To address this anti-jamming problem, a Markov decision process (MDP) with a two-dimensional action space consisting of transmit frequency and power is proposed in this paper, modeling the interaction between a normal communication link and the presence of malicious jammers in a frequency hopping (FH) communication system. Furthermore, we also prove the existence of the deterministic optimal policy of the proposed model theoretically. To obtain a policy for the communication link to avoid being jammed, the Dual Action Network-Based Deep Reinforcement Learning Algorithm, and Action Feedback Mechanism are proposed. The energy consumption and frequency switching overhead are considered and evaluated in both the proposed model and the algorithm. Finally, the proposed model and algorithm are verified not only in a virtual simulation environment but also in the field testing environment. The result suggests that the proposed algorithm is of great practical value for solving anti-jamming problems.

Filter bank multicarrier with offset quadrature amplitude modulation (FBMC‐OQAM) has gained significant importance as a contender for the fifth generation mobile communication system and also meets the application requirements of the Internet of things. The peak‐to‐average power ratio (PAPR) of the FBMC‐OQAM is very high due to using a same prototype filter for each subcarrier and the addition of different subcarriers at the transmitter. Therefore, a novel randomly assigning prototype filter (RAPF) method is proposed to reduce PAPR in FBMC‐OQAM system. Based on the proposed RAPF method, the authors analyse the performances of both PAPR and bit error rate (BER) in the FBMC‐OQAM system. Simulation results show that the proposed RAPF method helps to significantly reduce PAPR of FBMC‐OQAM system. Meanwhile, compared with conventional FBMC‐OQAM system, the proposed RAPF‐FBMC‐OQAM scheme has almost the same BER performance in the additive white Gaussian noise channel but is descended slightly only in the high signal‐to‐noise ratio region of the frequency selective fading channel.

The increasing demand of computation capacity has made many-core parallel processing (MPP) a compelling choice for computation-intensive applications. The networks-on-chip (NoC) architecture is an effective way to interconnect dozens of processing cores, while the logic circuits and the actual performance need to be verified in specific platform. We proposed and implemented the MACRON platform to provide verification for complicated applications based on NoC architecture by coordinating the software tool and the hardware devices closely. In MACRON, the virtual output queue with look-ahead routing is proposed to reduce the transmission delay through the NoC router. The heterogeneous processing elements: vector processor core, scalar processor core and accelerator core are designed, thus a thorough 'soft' signal processing can be approached. A real-time 4G wireless communication system based on NoC is demonstrated on this MACRON platform.

Water‐Filling (WF) is widely applied in power allocation in multichannel wireless communications. By mathematically linearising the optimal WF expression, the authors observe an intrinsic parallel‐shift property of WF, based on which a fast and efficient WF algorithm is proposed. Compared with the conventional WF algorithms, it greatly simplifies WF execution by removing the Lagrange multiplier (or water‐level) searching process. The authors further apply the proposed parallel‐shift WF to solve the power allocation problem in orthogonal frequency division multiplexing (OFDM)‐based underlay cognitive radios (CRs), with the objective of maximising the secondary user's throughput over all OFDM sub‐channels under the transmit power and the interference constraints. To this end, the existing algorithm adopts an iterative binary searching process to find the solution, where the conventional WF algorithm with Lagrange multiplier searching is invoked in each iteration. In contrast, the authors propose a new power allocation algorithm to remove the iterative binary searching process. It runs the simplified parallel‐shift WF only once, and then directly calculates the final solution using a power adjustment process (with the parallel‐shift property as the underlying enabling mechanism). Numerical results show that both the proposed parallel‐shift WF and the OFDM‐based CR power allocation algorithms can run multiple times faster than the existing counterparts, and the gap on the running time increases with the total number of OFDM sub‐channels in the system.

We consider power allocation in energy efficient OFDM (Orthogonal Frequency Division Multiplexing) or MIMO (Multiple-Input Multiple-Output) transmissions. An energy-per-goodbit (EPG) metric is used to gauge the average energy consumed for transmitting each bit. Existing works minimize EPG by searching for a set of dual variables which are used to compute the optimal power allocation in a Water-Filling (WF) expression. The searching process is complex and requires a long running time. In this paper, we propose a new algorithm to compute the optimal power allocation without searching any dual variables. Our algorithm is based on an iterative calculation of the total transmission power and it converges when the minimum EPG is reached. It takes WF as the basic building block. Unlike the conventional WF approach which needs to search for a Lagrange multiplier (i.e., the water level), we additionally propose a novel WF algorithm without searching the Lagarange multiplier. Numerical results show that our power allocation algorithm for EPG minimization, together with the embedded fast WF algorithm, can run multiple times faster than the existing ones.

In LTE-A (3GPP LTE-Advance) systems, CoMP (Cooperative Multi-Point) is adopted to enhance the performance of edge users. To maximize the edge user throughput, it is very crucial to properly determine the set of simultaneously served users in the same PRB (physical resource block) and cooperating BSs (base stations) for each selected user, as well as the transmit power of the BSs. In this paper, we first propose a simple scheduling algorithm to choose cell-edge mobile stations (MSs) and cooperating BSs for each PRB according to the RSRP (reference signal receiving power) of each MS, based on which the classic Water-Filling (WF) is applied at each BS to allocate transmit power over all PRBs. However, the objective of single cell power allocation is to maximize the throughput of each individual cell without considering interference among different cooperating BS sets. Therefore, we further formulate a power reallocation mechanism using non-cooperative game theory to refine the single cell WF result for interference coordination, which maximizes the total edge user throughput over all BSs and PRBs by properly taking CCI (co-channel interference) into account. Based on proving the existence of a unique Nash Equilibrium for the formulated game, we design an algorithm to find the Nash Equilibrium and demonstrate the performance gain through extensive simulation studies.

Under a given modulation scheme and a target BER (bit error rate) requirement, it is difficult to increase the number of bits transmitted in each symbol (i.e., the rate), since this generally requires a significant increase of transmit power. In this paper, we study how to increase the rate in multiuser multiple-input multiple-output (MU-MIMO) broadcast channels using only bit and power loading, without requiring additional transmit power. To solve this challenging problem, our approach is to carry out a joint design of bit loading and power allocation over all streams, and meanwhile take user and stream selection into account. Specifically, we maximize the total number of bits per symbol (rather than the Shannon capacity) over all selected streams, under a target BER constraint for each stream and a constant total transmit power constraint. To enable flexible user and stream selection as well as achieving a high total rate, we consider OSDM (orthogonal space division multiplexing) rather than BD (block diagonalization) based MU-MIMO broadcast channels. Numerical results show that our proposed joint bit and power loading with OSDM user and stream selection is effective in increasing the total transmission rate without consuming additional resources.

We consider energy detection based spectrum sensing for opportunistic SU (Secondary User) transmissions in cognitive radio networks. Due to the time-varying nature of wireless fading channels and PU (Primary User) activities, the instantaneous SINR (Signal to Interference plus Noise Ratio) at the SU receiver changes from slot to slot in a time-slotted system. Unlike the conventional energy detector which uses a fixed value of energy threshold to detect the PU's occurrence, we let the SU transmitter dynamically adjust the threshold according to the instantaneous SINR. Under the constraint of limiting the average interference to the PU within a target level, the objective is to maximize the SU's average transmission rate and throughput. Our task is to determine a proper policy function for threshold control, which formulates the value of the threshold as a function of the SINR to achieve the above objective. In particular, we consider a linear policy function, which allows a higher threshold and thus more aggressive SU transmissions under a larger SINR. Simulation results show that the SU's average transmission rate can be significantly improved using the optimized policy function.

We consider power allocation in OFDM based underlay cognitive radio networks with partially known inter-system CSI (Channel State Information). Under a given total transmit power limit at the SU (Secondary User) transmitter, the goal is to assign a certain amount of power for signal transmission in each OFDM sub-channel, such that the SU's overall throughput can be maximized, and the average interference to the PU can be kept within a target outage probability level. The existing algorithm adopts an iterative binary searching process to find the solution, where the classic Water-Filling (WF) is invoked in each loop, which needs a relatively long running time. In this paper, an efficient algorithm is proposed to solve the problem in a much simpler and faster way. Specifically, we first propose an efficient approach to implement WF based on some in-depth theoretical analysis. Then, a novel power allocation algorithm is proposed by removing the binary searching process. Our algorithm runs WF only once and then directly calculates the final solution. Numerical results show that it can run tens to hundreds times faster than the existing algorithms, depending on the total number of OFDM sub-channels.