On robust dirty paper coding

A dirty paper channel is considered, where the transmitter knows the interference sequence up to a constant multiplicative factor, known only to the receiver. We derive lower bounds on the achievable rate of communication by proposing a coding scheme that partially compensates for the imprecise channel knowledge.We focus on a communication scenario where the Gaussian noise is small while the interference is strong. Our approach is based on analyzing the performance achievable using extended Tomlinson-Harashima like coding schemes. When the power of the interference is finite, we show that this may be achieved by a judicious choice of the scaling parameter at the receiver. We further show that the communication rate may be improved, for finite as well as infinite interference power, by allowing randomized scaling at the transmitter.