Particle Filtering for Joint Symbol Detection, Frequency Offset and Channel Estimation in Time-Varying MIMO Channels with Multiple Frequency Offsets
- 72 Downloads
We develop three methods for the joint symbol detection, frequency offset and channel estimation in time-varying multiple-input multiple-output channels with multiple frequency offsets between transmit and receive antennas. These methods are based on particle filtering. The first method utilizes the posterior proposal distribution (PD) to generate particles, which is optimal PD because it minimizes the variance of the importance weights, conditionally on the observations and past particles. Second, we develop an improved sampling strategy, which exploits the discrete nature of the symbol variable. The improved sampling strategy has same computational complexity as the posterior PD, while its performance is significantly improved. Finally, we derive a suboptimal complexity-reduced method, which utilizes the artificial sequential structure of the Bell-Labs layered space–time detection scheme to compress the sample space of symbol variable. Compared to the posterior PD, the computational complexity of the suboptimal method is largely reduced, while it still significantly outperforms the posterior PD. Simulation results are provided to illustrate the performance of these methods.
KeywordsChannel estimation Frequency offset Multi-input multi-output (MIMO) Particle filter Symbol detection
- 23.Jakes, W. C. (1974). Microwave mobile communications. New York: Wiley.Google Scholar