Graz University of Technology
This paper presents the capacity-achieving input distribution of an energy detection receiver structure. A proper statistical model is introduced which makes it possible to treat the energy detector as a constrained continuous communication channel. To solve this non-linear optimization the authors used the Blahut-Arimoto algorithm extended with a particle method, so that also continuous channels can be handled. To get a better convergence behavior of the algorithm, they also implement two new methods, which are called "Fuse particles" and "Kick particles". The results they present show that the capacity of the energy detector decreases with increasing integration time and decreasing peakto-average power ratio.