With blind detection, one can enhance the throughput of communications. However, not all blind algorithms can be applied in time-varying or frequency-selective fading channels. Furthermore, blind algorithms usually have higher complexity. In this paper, we first propose a general principle by which one can build a metric for blind detection from any pilot-assisted channel response (CR) estimation algorithm. By this principle and based on the least-squares-fitting (LSF) pilot-assisted CR estimation, we derive a metric for blind detection and propose an efficient blind data detection algorithm. The proposed algorithm is operated on a block of received symbols that are along subchannels or along time slots; therefore, the proposed blind detection can be applied in frequency-selective or time-varying fading channels. With the proposed tree search method, the proposed algorithm attains the same error performance as a previous blind detection algorithm based on the same metric, although the complexity is greatly reduced. We further apply the channel prediction to obtain a better initial block in the tree search, such that the complexity can be further reduced. A range-reduced scheme is also proposed to give a tradeoff between performance and complexity. In the fast time-varying channel, we can combine the proposed blind detection with an intersubchannel interference (ICI) self-reduction algorithm.
All Science Journal Classification (ASJC) codes