Collapsed VBI-DP Based Structured Sparse Channel Estimation Algorithm for Massive MIMO-OFDM

Abstract

Massive multiple input multiple output (MIMO) technology significantly improves the capacity of wireless communication systems by deploying hundreds of antennas at the base station. However, the large scale of the array implies higher computational complexity and pilot overhead when implementing channel estimation in the uplink. Utilizing the sparse channel structure is a promising approach to improve the channel estimation performance while circumventing such problems. In this paper, we investigate the detailed physical structure in the delay-spatial domain of uplink channels in massive MIMO-orthogonal frequency division multiplexing (MIMO-OFDM) systems and construct a hierarchical probabilistic model based on Dirichlet process (DP) prior to match the channel's structural sparse features. Based on the model, we derive a structured sparse channel estimation algorithm by implementing collapsed variational Bayesian inference (CVBI). The simulation results demonstrate that the proposed CVBI-DP algorithm can improve channel estimation performance significantly compared with the state-of-the-art methods for massive MIMO-OFDM, without increasing the computational complexity and pilot overhead. INDEX TERMS Massive MIMO, structured sparse channel, Dirichlet process, collapsed variational Bayesian inference.

MIMO–OFDM communication systems for traffic data transmission in 5G drone small cells

“…One hundred frames were chosen from an actual video [14] that was taken by a drone, monitoring traffics. The frames were chosen from different drone altitudes and their direction towards the traffic.…”

“…For the MIMO-linearising-PEP method, we have chosen Q LP = 6 × KN a = 360 because we have the 3 × KN a < L ≤ 4 × KN a condition. It is indicated in [14] that the duration of fixed channel tap positions is approximately equal to f c T s times of the coherence time. Therefore, we have chosen N r = 3000 in our simulations.…”

“…There are several papers that proposed CS-based MIMO-OFDM CE methods [5]. Some of these papers study zero Doppler shift conditions and therefore, they proposed CE for frequency selective channels [6][7][8][9][10][11][12][13][14][15][16][17]. The previous CS-based CE methods for doubly selective (DS) channels, time and frequency selective, in MIMO-OFDM systems utilise scattered pilots in the frequency domain and can be categorised into three major groups.…”

Combined UAMP and MF Message Passing Algorithm for Multi-Target Wideband DOA Estimation with Dirichlet Process Prior

“…Many studies have proposed probabilistic models based on Dirichlet process (DP) prior to cluster different subbands in order to improve the accuracy of wideband DOA estimation [34][35][36][37] . Wang et al [34] and Lu et al [35] mainly inferred the hidden variables on mean field (MF) theory under the Bayesian networks.…”

“…Many studies have proposed probabilistic models based on Dirichlet process (DP) prior to cluster different subbands in order to improve the accuracy of wideband DOA estimation [34][35][36][37] . Wang et al [34] and Lu et al [35] mainly inferred the hidden variables on mean field (MF) theory under the Bayesian networks. Lu et al [36] proposed a combined belief propagation (BP) and MF message passing algorithm which can compute exact marginals to improve the channel estimation performance.…”