In this paper, we investigate an artificial intelligence and machine learning (AI/ML) aided channel state information (CSI) feedback scheme for beamforming transmission in next generation wireless local area networks (WLANs). We exploit the deep neural network autoencoder (DNN-AE) with an objective of minimizing CSI feedback overhead while maintaining packet error rate (PER) performance. The key component of the proposed DNN-AE is a pre-processor with a mathematical closed-form, which enables a single generalized DNN-AE applicable universally to various combinations of CSI types, bandwidth, transmit and receive antenna numbers. The generalized DNN-AE has benefits of not only reducing complexity required for hardware implementation but also alleviating the engineering efforts to train DNN-AE. Simulation results show the generalized DNN-AE can provide an average of 50% CSI feedback overhead reduction with almost same packet error rate (PER) performance, compared with existing Extremely High Throughput (EHT) IEEE 802.11be WLANs. This leads to throughput increase by 30%, which can satisfy the goal of next generation Ultra High Reliability (UHR) IEEE 802.11bn WLANs.

To support higher throughput in next-generation Wi-Fi systems, efficient compression and feedback of channel state information (CSI) from a station (STA) to an access point (AP) is essential. This paper proposes a deep learning (DL)-based CSI feedback framework tailored for Wi-Fi systems. The framework employs encoder and decoder networks to compress and reconstruct CSI angle parameters, with a trainable vector quantization (VQ) module enabling efficient finite-bit representation through end-to-end training. To further enhance performance, we introduce an angle-difference feedback strategy that exploits the temporal correlation of the angle parameters by feeding back the difference between the current and previous values. This is complemented by preprocessing that handles the periodicity of angles and tailored VQ modules that compensate for residual quantization errors. Additionally, we present a DL-based CSI refinement module at the AP, which improves reconstruction by jointly using current and prior feedback. Simulation results show that the proposed framework outperforms both standard Wi-Fi feedback and existing DL-based feedback methods, with notable gains from both angle-difference feedback and CSI refinement.

Channel smoothing is widely adopted in wireless local area network (WLAN) systems to improve channel estimation, but the combination of channel smoothing and beamforming has been still a challenging work so far. The difficulty of this combination is due to the existence of discontinuities in the beamforming matrices across subcarriers. We first propose a receiver design for channel state information (CSI) feedback with a smooth beamforming matrix in an optimal way. This design is based on the optimization to maximize the cross-correlation between adjacent beamforming matrices at a cost of slightly increased feedback overhead. Then, a sub-optimal design is also proposed which has the same amount of feedback overhead as current WLANs. To further reduce feedback overhead with minimum loss of performance, machine learning (ML) technique is adopted in which the CSI quantization level is determined dynamically based on real-time channel frequency selectivity. The simulation through the IEEE 802.11be link-level simulator shows that the proposed schemes can achieve channel smoothing gain in the beamforming, improving the throughput significantly with reduced feedback overhead.

Givens rotation based channel state information (CSI) feedback has been adopted as a limited feedback technique for beamforming in the wireless local area networks (WLANs). On the other hand, the long term evolution (LTE) systems utilize the predefined codebook for the CSI feedback. In this paper, we propose a dual CSI feedback technique for next generation WLANs, which combines the codebook and Givens rotation to derive the benefits of both techniques. Machine learning (ML) technique is adopted for the improved codebook design. The extensive simulation is carried out via IEEE 802.11be link-level simulator to verify the performance of the proposed scheme. It shows that the proposed scheme can reduce feedback overhead by more than 50% compared to the scheme adopted in current WLANs, and enhances the throughput significantly.

In this paper, we address a joint design of symbol level beam selection and precoding in the multi-user downlink beamspace multiple-input single-output (MISO) system. Unlike the general concept of eliminating or avoiding inter-user interference, the symbol level design aims to exploit the constructive interference at symbol level among the multiple users. To optimize the constructive interference while ensuring the signal interference noise ratio (SINR), a cardinality constrained minimum mean square error (MMSE) problem with per-beam power limitation is formulated. To resolve this non-convex and non-smooth problem, we propose a penalty proximal alternating linearized minimization based algorithm, in which a sequence of penalty sub-problems are solved by iteratively executing two projections: selective constructive signal region projection and sparse beam selection projection. From the extensive simulation results, the convergence of the proposed algorithm and its effectiveness are demonstrated.

Polar code is adapted as the channel coding scheme for control channel in 5G NR. The basic algorithm of polar code is the successive cancelation (SC) decoding but its performance is not good enough and decoding latency is relatively high. So the simplified successive cancelation list(SSCL) decoding is generally used. When the SSCL is operated, path metric calculator generates children candidates and the number of children candidates influenced on the H/W complexity. The reliability of children candidates is affected by that of their parent candidates. The number of children candidates is adaptively selected according to the reliability of their parent candidate and the total number of children candidates could be reduced. Our proposed algorithm reduces H/W complexity of polar decoder while preserving decoding performance.

Due to the short wavelength of mmWave channel in 5G, the size of antenna array becomes reduced and beamforming has been thus adopted in mobile handsets. To enable the beamforming, the phase shifter is necessary to control the phases from different antenna elements. However the phase shifter can have non-ideal phase and non-identical gain responses over the antenna elements and phase codes. To consider such impairments in practice, we suggest the code search algorithm using low-complexity measurement for desired beam directions and its applications in beam codebook generation. Using the numerical simulation, it is shown that the spherical coverage can be improved by the proposed procedure using the combination of modelling and measurements.

In this paper, a supervised-learning based blind detection of interference parameters for network-assisted interference cancellation and suppression (NAICS) system is proposed. In order to enable joint detection or interference cancellation at a user equipment in NAICS system, we consider detection of the interference parameters including traffic-to-pilot power ratio (TPR), rank indicator (RI), and precoding matrix indicator (PMI). We divide overall process for detection of these interference parameters into two steps, and propose supervised learning based neural network architecture for detection of corresponding parameters in each steps. Link-level simulation results are provided to validate detection performance of the proposed neural network architecture and performance of NAICS system which uses the proposed learning based detection method with respect to block error rate (BLER).

The exponential growth of data rate in advanced wireless communication systems results in an overwhelming overhead on physical links that transport complex-value digital signal between Radio Frequency (RF) unit and Baseband Processor (BBP) at mobile terminal. To suppress this overhead, we present the data compression which lowers the bit-width of inphase (I) and quadrature(Q) sample to be transferred over RF-BBP link. The proposed compression schemes are based on floating-point transform and bit-level modification targeting for orthogonal frequency division multiplexing (OFDM) signal in Long Term Evolution (LTE) and New Radio (NR) systems, which lead to cost-effective implementation with a low latency and reduced computational complexity. Through theoretical analysis and performance evaluation, it is verified that the intended data rate using 4 × 4 MIMO and 256QAM is achieved with an ignorable performance loss compared to the uncompressed mode, providing 25% compression ratio.

We propose a high-resolution hierarchical beam alignment algorithm that learns millimeter wave channel space using segmented beams obtained from feasible phase shifted sum of given hierarchical basis. To be specific, at each stage, the segmented beam is designed using a linear combination of given training beam basis instead of just choosing a beam among them. Then, the designed beam is used to determine the subset of higher resolution training beam candidates for the next stage which is referred to as the over-complete dictionary embracing even partially overlapped beams as elements. At the next stage, the proposed beam alignment can focus only on the most promising directions associated with finer grids. The simulation results demonstrate that the proposed beam alignment framework can remedy the limitations of the existing grid-of-beams method and achieve favorable beam alignment performance with reduced training time.