Enhanced data detection in OFDM systems using factor graph
In OFDM systems, the cyclic prefix (CP) eliminates the interblock interference and enables the use of the computationally efficient single-tap equalizers at the receiver. While posing a loss in both spectrum efficiency and power efficiency, the CP brings extra information about the data which can be used for detection. Therefore, instead of discarding the CP as in the conventional OFDM systems, this paper takes advantage of the prefix redundancy and utilizes it in the soft-input soft-output equalizer of a turbo equalization system. By using factor graph, an equalization algorithm is developed, and with proper approximation, the complexity of the proposed algorithm is reduced to O(2Rlog2N + 4RG/N log2G + 2RG/N) per data symbol for R iterations, where N is the length of the block and G is equal to P + L - 1 with P the CP length and L the channel length. Simulation results show that the turbo equalization system converges within two iterations and the proposed equalization approach achieves significant gain compared to the conventional approach.