EXIT-chart-based power allocations for superposition multilevel coding systems

Multilevel coding with sigma-mapping is one simple way to obtain coding gain as well as shaping gain, while power-allocation, by allocating proper powers to different levels, can facilitate the iterative decoding/detection algorithms. In this paper, we utilize the EXIT chart to calculate the convergence threshold for each level by treating the signals from other levels as Gaussian noise. Based on the threshold, we propose a simple power allocation scheme. Numerical results show that the proposed EXIT-based power allocation scheme improves the performance when compared with the existing Gaussian approximation power allocation scheme while still keeping a reasonable complexity.