On the information rate of sparse ISI channels

Sparse ISI channels have been considered for applications in underwater acoustic, wireless multipath, aeronautical, and satellite systems. They are characterized by having a long gain vector with few significant taps. BCJR-based simulation methods used when the channel memory is low become quickly impractical since the number of states grows exponentially large with it. A novel techniques is proposed in this paper based on simple information theoretical considerations and counting arguments, which is applied to the i.i.d. binary antipodal input case. Simulation results are reported illustrating how the information rate can be determined for either short ISI channels and long sparse ISI channels with 50-symbol memory.