Variable Is Better Than Invariable: Sparse VSS-NLMS Algorithms with Application to Adaptive MIMO Channel Estimation

Table 1

Simulation parameters.

Parameters	Values
Number of transceivers (𝑁𝑡,𝑁𝑟)	(4,4)
Channel length of each 𝐡𝑛𝑟𝑛𝑡	𝐿=16
Number of nonzero coefficients	𝑇∈{1,4}
Distribution of nonzero coefficient	Random Gaussian 𝒞𝒩(0,1)
Threshold parameter for VSS-NLMS	𝐶∈{10^−4,10^−5}
Received SNR for channel estimation	{5 dB, 15 dB, 25 dB}
Received SNR 𝐸0/𝑁0 for data trans.	12 dB~30 dB
Step-size	𝜇=0.2 and 𝜇max=2
Regularization parameter for ZA-NLMS	𝜌ZA=0.0006𝜎^2𝑛 for 𝑇=1; 𝜌ZA=0.0002𝜎^2𝑛 for 𝑇=4
Regularization parameter for RZA-NLMS	𝜌RZA=0.006𝜎^2𝑛 for 𝑇=1; 𝜌RZA=0.002𝜎^2𝑛 for 𝑇=4
Modulation schemes	QPSK, 8PSK, 16PSK, 16QAM, 64QAM, 128QAM