This paper proposes a deep learning based power allocation (DL-PA) and hybrid precoding technique for multiuser massive multiple-input multiple-output (MU-mMIMO) systems. We first utilize an angular-based hybrid precoding technique for reducing the number of RF chains and channel estimation overhead. Then, we develop the DL-PA algorithm via a fully-connected deep neural network (DNN). DL-PA has two phases: (i) offline supervised learning with the optimal allocated powers obtained by particle swarm optimization based PA (PSO-PA) algorithm, (ii) online power prediction by the trained DNN. In comparison to the computationally expensive PSO-PA, it is shown that DL-PA greatly reduces the runtime by 98.6%-99.9%, while closely achieving the optimal sum-rate capacity. It makes DL-PA a promising algorithm for the real-time online applications in MU-mMIMO systems.
This paper investigates a multi-user massive multiple-input multiple-output (MU-mMIMO) hybrid precoding (HP) scheme using low-resolution phase shifters (PSs) and digital-to-analog converters (DACs). The proposed HP approach involves two stages: RF beamforming based on the slowly time-varying channel second-order correlation matrix, and baseband MU precoding based on the instantaneous effective baseband channel to mitigate MU-interference by a regularized zero-forcing (RZF) technique. We consider three HP design architectures: (i) HP using full-resolution PSs and DACs, with a baseband transfer block for constant-modulus RF beamformer, (ii) HP using b-bit PSs and full-resolution DACs, with an orthogonal matching pursuit (OMP) based algorithm that can approach the optimal unconstrained RF beamformer, and (iii) HP using b-bit PSs and q-bit DACs, taking into account also DAC quantization noise. Illustrative results show that the proposed HP schemes with low-resolution PSs can approach the sum-rate of full-resolution PSs by using only 2-bit PSs, while offering higher energy efficiency. Furthermore, a study of sum-rate results for various PS and DAC quantization levels reveals that HP can achieve near-optimal performance with only 2-bit PSs and 5-bit DACs. Moreover, a comparison of the different array configurations, namely, uniform linear array (ULA), uniform circular array (UCA), uniform rectangular array (URA), and concentric circular array (CCA), indicates that URA and CCA outperform UCA and ULA in terms of spectral and energy efficiencies.
This work considers the relay selection and resource allocation problem (i.e., link scheduling, and rate allocation) for multi-source, multi-relay dual-hop wireless networks. The relays employ buffers to store the received data from the sources for future transmissions. End-to-end (E2E) delay of each traffic flow originated from a source or a relay is constrained in terms of maximum allowable delay-outage probability. To solve this problem, we first study the resource allocation problem to maximize the constant supportable arrival rate of a non-prioritized source under minimum rate requirements of the prioritized sources and relays for a given relay selection solution. Then, the optimal relay selection can be determined to support the largest rate of the non-prioritized source among all possible relay selection solutions. We derive the resource allocation solutions using asymptotic delay analysis and convex optimization techniques. We also develop an online allocation algorithm which does not require the knowledge of the fading statistics by using stochastic approximation theory. Numerical results are presented to demonstrate the usefulness of the proposed resource allocation design for relay selection under different delay and rate constraint regimes.
Zero-forcing (ZF) precoding scheme can achieve the asymptotic sum capacity as dirty-paper coding (DPC) in multiple-input multiple-output broadcast (MIMO-BC) channel when the number of users, , approaches infinity. However, the gap between ZF and DPC is not negligible in a practical range of , that is, . The capacity loss is partly due to the excessive transmission power penalty incurred by ZF when the channel matrix of the selected user subset is poorly conditioned. To avoid this power penalty, we propose to use a variation of ZF, channel inversion regularization (CIR), as a precoding scheme in MIMO-BC channels. But, unlike the interference-free ZF, the problem of maximizing sum-rate capacity using CIR precoding becomes nonconvex, which cannot be solved by water-filling strategy. Thus, we propose an efficient algorithm based on gradient projection (GP) as the optimal power allocation strategy for selected users, and show that the proposed CIR precoding scheme can achieve asymptotically the optimum sum-rate of the DPC strategy. Moreover, simulation results show that the CIR precoding scheme with the proposed optimal power allocation scheme achieves better sum-rate performance than ZF for a wide range of .
Two interference-based sub-carrier group assignment strategies in dynamic resource allocation are proposed for MC-CDMA wireless systems to achieve high throughput in a multi-cell environment. Least interfered group assignment (LIGA) selects for each session the sub-carrier group on which the user receives the minimum interference, while best channel ratio group assignment (BCRGA) chooses the sub-carrier group with the largest channel response-to-interference ratio. Both analytical framework and simulation model are developed for evaluation of throughput distribution of the proposed LIGA and BCRGA schemes. An iterative approach is devised to handle the complex interdependency between multi-cell interference profiles in the throughput analysis. Illustrative results show significant throughput improvement offered by the proposed interference-based assignment schemes for MC-CDMA multi-cell wireless systems. In particular, under low loading conditions, LIGA renders the best performance. However, as the load increases BCRGA tends to offer superior performance.
Multiple-input-multiple-output (MIMO) precoder design for frequency-selective fading channels using partial channel information based on the spatial and path correlation matrices is presented. By representing a frequency-selective fading channel as a multipath model with L effective paths, a general precoding structure is proposed and used to derive optimum precoding designs that maximize Jensen's upper bound on the channel ergodic capacity under the transmitted power constraint for two cases, i.e., uncorrelated and correlated channel paths. Analytical results show that, in the uncorrelated case, the precoder structure consists of a number of parallel precoders for frequency-flat fading channels. The power assignment to each precoder and the power allocation over the eigenmodes of each precoder are calculated based on the power of channel paths and the eigenvalues of the transmit correlation matrix. In the correlated case, the precoder structure is an eigenbeamformer with the beams referred to a function of eigenvectors of the Kronecker product of path and transmit correlation matrices. Furthermore, the power allocated to each eigenmode can be obtained from a statistical water-pouring policy that is specified by the product of eigenvalues of the transmit antenna and path correlation matrices. Simulation results for different scenarios indicate that the proposed precoder can increase the ergodic capacity of MIMO systems in a frequency-selective fading environment with spatial and path correlations, and its offered capacity gain is increased with the level of correlation and numbers of antennas and channel paths.
This paper examines a CoMP system where multiple base-stations (BS) employ coordinated beamforming to serve multiple mobile-stations (MS). Under the dynamic point selection mode, each MS can be assigned to only one BS at any time. This work then presents a solution framework to optimize the BS associations and coordinated beamformers for all MSs. With target signal-to-interference-plus-noise ratios at the MSs, the design objective is to minimize either the weighted sum transmit power or the per-BS transmit power margin. Since the original optimization problems contain binary variables indicating the BS associations, finding their optimal solutions is a challenging task. To circumvent this difficulty, we first relax the original problems into new optimization problems by expanding their constraint sets. Based on the nonconvex quadratic constrained quadratic programming framework, we show that these relaxed problems can be solved optimally. Interestingly, with the first design objective, the obtained solution from the relaxed problem is also optimal to the original problem. With the second design objective, a suboptimal solution to the original problem is then proposed, based on the obtained solution from the relaxed problem. Simulation results show that the resulting jointly optimal BS association and beamforming design significantly outperforms fixed BS association schemes.
Error-trapping decoding techniques are attractive due to their simple structure. Since 1962 several improved error-trapping methods have been devised in an effort to extend the capability and effectiveness in decoding multiple-error-correcting cyclic codes. Prange (1962) and MacWilliams (1964) introduced a (T, U) permutation group applied to this error-trapping decoding strategy by making use of a set of code-preserving permutation to obtain k error-free positions from which the rest of the code word could be reconstructed. Recently, exact lower bounds on the code length n for (n, k, 2t+1) cyclic codes have been found by using 5-step and 3-step (T, U) permutation groups. The present paper presents a study on the relationship between the code parameters n, k, t and the number of permutation steps s, with t being odd. Some examples on the capability of (T, U) permutation decodable (PD) cyclic codes are illustrated.< >
A decode-and-forward cooperative relaying scheme where an N T -antenna source is assisted by several N D -antenna relays to forward its information to a N D -antenna destination is considered. Its symbol error rate (SER) and average transmit power per transmit antenna (ATPT) are derived and validated by computer simulations. Illustrative results over different network topologies show that the considered cooperative relaying scheme offers a significant performance improvement as compared to direct transmission and cooperative relaying without direct link under the same transmission power and bandwidth efficiency.
We consider a three-node buffer-aided relaying network with statistical quality-of-service (QoS) constraint in terms of maximum acceptable end-to-end queue-length bound outage probability. In particular, we study the adaptive link selection relaying problem that aims to maximize the constant supportable arrival rate μ to the source (i.e., the effective capacity). Fixed and adaptive source and relay power allocation are investigated. By employing asymptotic delay analysis, we first convert the QoS constraint into minimum QoS exponent constraints at the source and relay queues. We then derive the link selection and power allocation solutions as functions of the instantaneous link conditions and QoS exponents using Lagrangian approach. Solutions for various special cases of link conditions and QoS constraints are presented. Moreover, we compare the effective capacities of the proposed relaying schemes and other existing schemes under different link conditions and QoS constraints. Illustrative results indicate that the proposed schemes offer substantial performance gains, and power adaption outperforms fixed power allocation at low signal-to-noise power ratio (SNR) region or under loose QoS constraints.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
'%3e%3cg id='b'%3e%3cpath id='a' stroke='%23000' stroke-width='2' d='M-6-25H6m-12 3H6m-12 3H6'/%3e%3cuse xlink:href='%23a' y='44'/%3e%3c/g%3e%3cpath stroke='%23fff' d='M0 17v10'/%3e%3ccircle r='12' fill='%23cd2e3a'/%3e%3cpath fill='%230047a0' d='M0-12A6 6 0 0 0 0 0a6 6 0 0 1 0 12 12 12 0 0 1 0-24z'/%3e%3c/g%3e%3cg transform='rotate(-123.69)'%3e%3cuse xlink:href='%23b'/%3e%3cpath stroke='%23fff' d='M0-23.5v3M0 17v3.5m0 3v3'/%3e%3c/g%3e%3c/svg%3e)