Achievable Rates Of Coordinated Multi-point Transmission In LTE-A System

In recent few years, coordinated multi-point transmission is proposed as a promising method,which can suppress inter-cell interference by coordinating among the multiple base stations. ForLTE-Advanced system, the increasing throughput of the system and the cell edge by usingcoordinated multi-point (CoMP) transmission is of special interest.In practical situations, channel state information (CSI) is required to make the schedulingdecisions at each base station. However, the channel state information is difficult to get and is neverperfect, such as limited feedback errors, estimation errors and feedback delay. As a result, it is morenecessary to analyze the achievable rates of different coordinated multi-point modes under theimperfect channel state information to promote the insight of the new technology.This paper studies the achievable rates of coordinated multi-point transmission under the perfectand imperfect channel state information in LTE-Advanced system. The achievable rates are derivedfor both joint processing mode and coordinated beamforming mode considering the limitedfeedback errors, estimation errors and feedback delay.Under the situation of the equal power allocation, the system throughput of joint processing (JP)mode is first provided by the probability distribution of SNR. Then, we propose two approximatemethods to simplify the calculation complexity, which is average-substituted anddistribution-equivalent respectively. Hence, we get the simple and accurate closed-form expressionsinstead of using brute force due to the complicated distribution. For the coordinated beamforming(CBF) mode, we derive the achievable rates based on the probability distributions of signal termand interference term. Furthermore, we propose a mode switching scheme to achieve the maximalsum throughput according to SNR and quantization accuracy.Under the situation of the unequal power allocation, we propose two approximate methods tocalculate the system throughput. The one is distribution-equivalent. It is common practice instatistics to approximate a weighted sum of chi-square variables by a single one with differentdegrees of freedom and an adequate scaling factor. The other is using the formula, so we can easilyget the probablility distribution function of SNR, which is the sum of chi-square variables with theunequal coefficients.Finally, numerical results show that the proposed methods provide performance close to thatachieved by the simulation results.