A conditional ℓ1 regularized MMSE channel estimation technique for IBI channels

Inter-block-interference (IBI) caused in the pursuit of spectral efficiency can deteriorate channel estimation performance. For this problem, previously-proposed chained turbo estimation performs IBI cancelation by using the soft replica of the transmitted signal. The IBI cancelation technique can, however, suffer from a mean squared error (MSE) floor problem, since the soft replica is unavailable at the first turbo iteration. The IBI problem can be avoided by using channel impulse response length constraint. Nevertheless, as shown in this paper, the IBI avoidance approach is difficult to perform independently since it requires unbiased second-order statistics. This paper proposes, therefore, a new conditional ℓ1 regularized minimum mean square error channel estimation algorithm by jointly utilizing the IBI avoidance/cancelation and subspace techniques. Simulation results verify that the proposed algorithm solves the MSE floor problem, and, hence, improves the bit error rate convergence performance in realistic IBI channels including the effect of pulse shaping filters.

Takano Yasuhiro, Su Hsuan-Jung, Juntti Markku, Matsumoto Tad

Publication type:
A1 Journal article – refereed

Place of publication:

1 norm regularization, compressive sensing; turbo equalization, inter-block-interference (IBI), subspace-based channel estimation


Full citation:
Y. Takano, H. Su, M. Juntti and T. Matsumoto, “A Conditional ℓ1 Regularized MMSE Channel Estimation Technique for IBI Channels,” in IEEE Transactions on Wireless Communications, vol. 17, no. 10, pp. 6720-6734, Oct. 2018. doi: 10.1109/TWC.2018.2863295


Read the publication here: