Optimization for training in multiple antenna wireless system with equal power collaboration

Multiple-antenna systems equipped with the channel estimation techniques can achieve very high data rate with low error probabilities, especially when the CSI (channel state information) is known at the receivers. It´s well accepted that sending the training sequence at the transmit antennas is an effective method to acquire the CSI. Optimized power collaboration can absolutely provide better capacity than equal power, but in some communication schemes there´s no necessity to pay the luxury of varying the power during transmission. This paper would show that the capacity difference between optimized and equal power is acceptable. Based on the lower capacity bound, this paper proposed a closed formula to estimate the optimized training length for equal power, which can avoid the huge amount of searching and computation brought by the Monte Carlo solution. The simulation results show that the training length computed by the formula can achieve almost the same capacity as the Monte Carlo solution.