A factor graph approach to joint channel estimation and symbol detection over fast flat fading channels

In this paper, a novel joint channel estimation and symbol detection method over fast flat fading channel is proposed, by applying sum-product algorithm on factor graph. An appropriate graph model is derived from factoring the likelihood function of transmitting sequence. In addition, particle filter is introduced to reduce the computational complexity of continuous integral within sum-product algorithm. This method provides an intuitive perspective on joint channel estimation and symbol detection, and improves spectrum utilization efficiency compared with conventional pilot-only technique.