CSI Feedback Reduction for MIMO Interference Alignment

A novel interference alignment (IA) scheme based on the errors-in-variables (EIV) mathematic model has been proposed to overcome the channel state information (CSI) estimation error for the MIMO interference channels. By solving an equivalently unconstrained optimization problem, the proposed IA scheme employing a weighted total least squares (WTLS) algorithm can obtain the solution to a constrained optimization problem for transmit precoding (TPC) matrices and minimizes the distortion caused by imperfect CSI according to the EIV model. It is shown that the design of TPC matrices can be realized through an efficient iterative algorithm. The convergence of the proposed scheme is presented as well. Simulation results show that the proposed IA scheme is robust over MIMO interference channels with imperfect CSI, which yields significantly better sum rate performance than the existing IA schemes such as distributed iterative IA, maximum signal-to-interference-plus-noise ratio (Max SINR), and minimum mean square error (MMSE) schemes.

Deep learning (DL)-based channel state information (CSI) feedback improves the capacity and energy efficiency of massive multiple-input multiple-output (MIMO) systems in frequency division duplexing mode. However, multiple neural networks with different lengths of feedback overhead are required by time-varying bandwidth resources. The storage space required at the user equipment (UE) and the base station (BS) for these models increases linearly with the number of models. In this paper, we propose a DL-based changeable-rate framework with novel quantization scheme to improve the efficiency and feasibility of CSI feedback systems. This framework can reutilize all the network layers to achieve overhead-changeable CSI feedback to optimize the storage efficiency at the UE and the BS sides. Designed quantizer in this framework can avoid the normalization and gradient problems faced by traditional quantization schemes. Specifically, we propose two DL-based changeable-rate CSI feedback networks CH- CsiNetPro and CH- DualNetSph by introducing a feedback overhead control unit. Then, a pluggable quantization block (PQB) is developed to further improve the encoding efficiency of CSI feedback in an end-to-end way. Compared with existing CSI feedback methods, the proposed framework saves the storage space by about 50% with changeable-rate scheme and improves the encoding efficiency with the quantization module.

Interference alignment (IA) is an effective method that can eliminate interferences in wireless networks, and has been applied to spectrum sharing in cognitive radio (CR) networks recently. However, the availability of perfect network channel state information (CSI) is necessary for most existing IA schemes, which is not practical in general due to the realistic communication scenarios and deployment challenges. In this paper, we apply IA to cognitive relay networks under CSI mismatch where the variance of the CSI measurement error depends on signal-to-noise ratio (SNR). An adaptive Max-SINR IA algorithm has been introduced to improve the performance of the secondary network by using the knowledge of CSI error variance. Finally, we analyze the performance of the secondary network in terms of the end-to-end equivalent transmission rate and outage probability. Simulation results indicate that our proposed adaptive Max-SINR IA scheme can greatly improve the performance of the secondary network.

We consider a downlink heterogeneous network (HetNet), where several picocells coexist in the same spectrum as a macrocell. For a general framework of the multiple-input multiple-output (MIMO) network consisting of arbitrary number of picocells and macro users within the macrocell, designing a closed-form solution to achieve the degrees of freedom (DoF) of (C + K)d while guaranteeing the minimum overall antenna usage is an open problem, where C, K and d are the number of picocells, marco users within the macro cell and data streams per user, respectively. Therefore, by relaxing the minimum antenna utilization constraint, we present a zero-forcing based transceiver design at first, which outperforms the existing closed-form solution based on interference alignment (IA) in terms of antenna usage in most cases. That is to say, alignment processing is not necessary in most cases for the existing scheme because simple zero-forcing scheme with efficient antenna utilization can also attain the same DoF. Besides, we propose two IA schemes without any iterative computations, which have the same antenna configuration but different CSI feedback overhead. Also, our proposed IA schemes require lower usage of antennas compared to the zero-forcing scheme and existing IA scheme.

In wireless interference networks, orthogonal medium access techniques like time division multiple access (TDMA) and frequency division multiple access (FDMA) achieve merely one degree of freedom (DoF). However, by using interference alignment (IA) the achievable DoF can be linearly scaled up with the number of users. Nevertheless, this outperformance of IA heavily depends on the availability of perfect channel state information (CSI), which is not a realistic assumption in practice. Since under imperfect CSI, the performance of IA may become severely degraded, design of enhanced IA algorithms by relying on the availability of merely imperfect CSI is of particular interest. In this paper, we propose a least squares (LS)-based IA algorithm which adaptively designs the beamformers based on the knowledge of the channel estimation error variance, which is possible to be known in advance. Then, we compare the performance of the proposed scheme with interference leakage minimization algorithms. It is shown that under both perfect and imperfect CSI, the proposed LS-based IA outperforms interference leakage minimization algorithms.

In this work we investigate the degrees of freedom (DoF) performance of the multiple-input multiple-output (MIMO) interference channel (IC) with limited channel state information (CSI) feedback. We generalise two space-time interference alignment (STIA) algorithms to the K user MIMO IC with a transmit power constraint and show how these schemes exploit both current and outdated CSI to achieve an improved DoF performance. We show that by combining the STIA schemes with traditional methods major DoF gain can be achieved for a variety of limited CSI feedback scenarios in comparison to what would be obtained via the use of the traditional techniques on their own. In particular we prove that optimal DoF can still be achieved in situations where current CSI is unavailable for a derived fraction of the time. Finally we validate the promised DoF results via system simulation and show how these DoF gains reflect in terms of sum rate achievable.

Aiming at the problem of high complexity and low feedback accuracy of existing channel state information (CSI) feedback algorithms for frequency-division duplexing (FDD) massive multiple-input multiple-output (MIMO) systems, this paper proposes a CSI compression feedback algorithm based on deep learning (DL), which is suitable for single-user and multi-user scenarios in massive MIMO systems. This algorithm considers the spatial correlation of massive MIMO channel and uses bidirectional long short-term memory (Bi-LSTM) and bidirectional convolutional long short-term memory (Bi-ConvLSTM) network to decompress and recover the CSI for single-user and multi-user, respectively. The proposed DL-based CSI feedback network is trained offline by massive MIMO channel data and could learn the structural characteristics of the massive MIMO channel by fully exploiting the channel information in the training samples. The simulation results show that compared with several classical CSI compression feedback algorithms, the proposed CSI compression feedback algorithm has lower computational complexity, higher feedback accuracy, and better system performance in massive MIMO systems.

A channel state information (CSI) feedback algorithm based on scalar quantization is proposed for a multiple-input multiple-output (MIMO) transmission. The performance of the proposed algorithm is evaluated using the effective transmission rate, which incorporates the effect of the CSI feedback overhead on transmission time. Besides that, the power consumed due to the computational complexity of the CSI feedback algorithm is taken into account together with the transmit power when evaluating the energy efficiency. We compare our proposed algorithm to other typical CSI feedback algorithms under correlated channel conditions. Simulation results show that the proposed algorithm enables practical MIMO transmission to achieve an relative transmission rate at 95% of the ideal rate.

In this paper, we propose a cross-attention and self-attention Transformer-based network (CASA-TransNet) for channel state information (CSI) feedback in dual-polarized massive multiple-input multiple-output (MIMO) systems. In dualpolarized massive MIMO systems, the channels in the horizontal and vertical polarization directions have a high correlation, which can be exploited for joint compression and reconstruction of the dual-polarized channels to improve the CSI feedback performance. The proposed CASA-TransNet employs a twostream cross-attention Transformer architecture and extracts the correlation between horizontal and vertical polarization CSI via cross-stream exchange of the query embeddings. Furthermore, a self-attention Transformer is utilized to model long-term dependencies of the dual-polarized CSI. Simulation results show that the proposed CASA-TransNet outperforms other deep learningbased methods, achieving better CSI reconstruction performance even at high compression ratios. Compared to state-of-the-art schemes, CASA-TransNet achieves a performance gain of $3.47 \sim 5.74 \mathrm{~dB}$.

Most learning-based channel state information (CSI) feedback efforts concentrate on enhancing feedback accuracy through innovative neural network (NN) designs and exploiting correlations. This paper introduces an integrated learning framework for CSI feedback and localization designed to synergistically improve both tasks. We present a novel unified approach for CSI feedback and downlink CSI-based localization, where feedback is facilitated by an autoencoder, and the downlink CSI-based localization uses the feedback codeword directly without requiring reconstruction. The goal is to simultaneously minimize feedback and localization errors. Additionally, for users with access to coarse position data, we propose a refined framework that integrates this information into both the feedback mechanism and localization processes. This coarse positional knowledge is incorporated into the encoding and decoding stages to reduce feedback errors and is inputted into the localization NN to enhance localization accuracy. The improved framework is refined through an end-to-end training strategy, focusing on concurrently reducing feedback and localization errors. Simulation results using ray tracing channel datasets demonstrate that our proposed method not only enables feedback and localization tasks to mutually benefit but also shows that incorporating coarse positional data significantly increases the accuracy of both CSI feedback and CSI-based localization.

For massive multiple-input multiple-output (MIMO) systems, downlink channel state information (CSI) feedback is critical for exploiting the performance potentials. Recognized by the 3rd Generation Partnership Project as a key use case for the artificial intelligence-native air interface, deep learning (DL)-based CSI feedback has attracted extensive research from academia and industry. However, one of the most significant obstacles to its practical deployment is the limited generalization capabilities of DL models. In this article, we first illustrate the generalization challenge faced by DL-based CSI feedback with the main factors. Then, we systematically analyze its feasible solutions from two perspectives: initial model acquisition and efficient online updating. Finally, to address knowledge-forgetting problems in online updating, a novel alternating optimization framework is proposed to further improve the generalization ability through knowledge review. Hopefully, this article can give inspiring insights for the optimization of deployment and generalization for DL-based CSI feedback systems.

The 3rd Generation Partnership Project began studying Release 18 in 2021. Artificial intelligence (AI)-native air interface is one of the key features of Release 18, where AI for channel state information (CSI) feedback enhancement is selected as the representative use case. This article provides an overview of AI for CSI feedback enhancement in 5G-Advanced. Several representative non-AI and AI-enabled CSI feedback frameworks are introduced and compared. Then, the standardization of AI for CSI feedback enhancement in 5G-advanced is presented in detail. First, the scope of the AI for CSI feedback enhancement in 5G-Advanced is presented and discussed. Then, the main challenges and open problems in the standardization of AI for CSI feedback enhancement, especially focusing on performance evaluation and the design of new protocols for AI-enabled CSI feedback, are identified and discussed. This article provides a guideline for the standardization study of AI-based CSI feedback enhancement. © 2002-2012 IEEE.

Large language models (LLMs) have revolutionized natural language processing by learning rich representations from massive and diverse text data. While pre-trained LLMs have been widely customized for a variety of text-related tasks, their potential in wireless communications remains largely untapped. In this paper, we propose a novel channel state information (CSI) feedback framework for massive multiple-input multiple-output (MIMO) systems that leverages pre-trained LLMs to bridge the gap between natural language processing and wireless communication. To adapt LLMs for the CSI feedback task, we design specialized pre-processing, embedding, and post-processing modules that convert CSI data into token representations compatible with LLMs. Moreover, to address the limitations of a single model and enhance performance in diverse environments and under various compression ratios, we introduce a mixture-of-agents (MoA)based framework that integrates the complementary strengths of multiple LLMs. Simulation results demonstrate that our approach achieves superior CSI reconstruction performance, robust generalization to unseen scenarios, and significant training cost reduction even with limited data. These findings confirm that the transferable knowledge acquired from large-scale natural language pre-training can effectively enhance CSI feedback, offering a promising direction for intelligent and efficient nextgeneration wireless communication systems.

In the presence of perfect channel state information (CSI), the achievable degrees of freedom (DoF) in wireless interference networks can be linearly scaled up with the number of users. Achievability is based on the idea of interference alignment (IA). However, in the presence of imperfect CSI, the sum rate becomes degraded, and full DoF may no longer be achievable. In this paper, the authors propose novel least squares (LS)- and minimum mean square error (MMSE)-based IA schemes that adaptively design beamformers by relying on the availability of imperfect CSI and knowledge of the channel estimation error variance in advance. Interestingly and unlike the other robust algorithms, the proposed adaptive schemes do not impose extra computational complexity compared to their nonadaptive counterparts. It is shown that the LS-based IA is able to outperform interference leakage minimization algorithms under both perfect and imperfect CSI. Furthermore, the authors compare the performance of the proposed MMSE-based IA with maximum signal-to-interference-plus-noise ratio (Max-SINR) algorithm. The authors show that while under perfect CSI, the MMSE-based IA achieves the same performance as that of Max-SINR, the former outperforms the latter under CSI mismatch. Meanwhile, it is shown that the proposed MMSE-based IA needs less CSI to be available and has less computational complexity compared to Max-SINR.

We show that the three-user M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> × M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> Multiple-Input Multiple-Output (MIMO) interference channel where each transmitter is equipped with M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> antennas and each receiver is equipped with M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> antennas has (M) d(M, N) = <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Δ</sup> min (M/2-1/κ, N/2+1/κ) degrees of freedom (DoF) normalized by time, frequency, and space dimensions, where = <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Δ</sup> min(M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> , M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> ), N = <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Δ</sup> max(M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> , M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> ), κ = <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Δ</sup> ⌈M/N-M⌉. While the DoF outer bound of d(M, N) is established for every M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> , M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> value, the achievability of d(M, N) DoF is established in general subject to a normalization with respect to spatial extensions, i.e., the scaling of the number of antennas at all nodes. In particular, we show that qd(M, N) DoF are achievable for the three-user qMT × qMR MIMO interference channel, for some positive integer q, which may be seen as a spatial extension factor. q is the scaling factor needed to make the value qd(M, N) an integer. Given spatial extensions, the achievability relies only on linear beamforming based interference alignment schemes and requires neither channel extensions nor channel variations in time or frequency. In the absence of spatial extensions, it is shown through examples how essentially the same interference alignment scheme may be applied over time-extensions over either constant or time-varying channels. The central new insight to emerge from this paper is the notion of subspace alignment chains as the DoF bottlenecks. The subspace alignment chains are instrumental both in identifying the extra dimensions to be provided by a genie to a receiver for the DoF outer bound, as well as in the construction of the optimal interference alignment schemes. The DoF value d(M, N) is a piecewise linear function of M, N, with either M or N being the bottleneck within each linear segment, whereas the other value contains some redundancy, i.e., it can be reduced without reducing the DoF. The corner points of these piecewise linear segments correspond to two sets, A={1/2, 2/3, 3/4, ...}and B={1/3, 3/5, 5/7, ...}. The set A contains all those values of M/N and only those values of M/N for which there is redundancy in both M and N, i.e., either can be reduced without reducing the DoF. The set B contains all those values of M/N and only those values of M/N for which there is no redundancy in either M or N, i.e., neither can be reduced without reducing the DoF. Because A and B represent settings with maximum and minimum redundancy, essentially they are the basis for the DoF outer bounds and inner bounds, respectively. Our results settle the question of feasibility of linear interference alignment, introduced previously by Cenk et al., for the three-user M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> × M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> MIMO interference channel, completely for all values of M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> , M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> . In particular, we show that the linear interference alignment problem (M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> × M <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</sub> , d) <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> (as defined in previous paper by Cenk et al.) is feasible if and only if d ≤ d(M, N). With the exception of the values M/N ∈ B, and only with that exception, we show that for every M/N value there are proper systems (as defined by Cenk et al.) that are not feasible. Evidently the redundancy contained in all other values of M/N manifests itself as superfluous variables that are not discounted in the definition of proper systems, thus creating a discrepancy between proper and feasible systems. Our results show that M/N ∈ A are the only values for which there is no DoF benefit of joint processing among co-located antennas at the transmitters or receivers. This may also be seen as a consequence of the maximum redundancy in the M/N ∈ A settings.