Research And Implement On Equalization Algorithm In Burst Communication Under Low SNR

In a digital communication system, an equalizer is used to eliminate the inter-symbol interference (ISI) of the received signal in the receiver, which is caused by the channel multipath propagation. Different from traditional communication, burst communication transmits information in a random and burst form. This kind of communication system ensures a strong concealment and anti-intercept ability with short duration and less amount of burst signal sequence, but at the same time, it causes a huge challenge to the equalizer. This paper studies and implements an equalization technique for burst signal with low signal-to-noise ratio (SNR).In this paper, we focus on equalization algorithms that minimize the system error rate using short training sequence in burst signal. We have studied the least mean square algorithm (LMS) based on the minimum mean square error criterion (MMSE), we have also studied both the least minimum bit error rate algorithm (LMBER) and the adaptive minimum bit error rate algorithm (AMBER) based on the minimum bit error rate criterion (MBER), which allowing us to figure out there performance in burst communication system.We have constructed a burst communication system model, compared and analyzed the error rate performance, convergence speed and computational complexity of LMS algorithm, LMBER algorithm and AMBER algorithm utilizing Matlab tool.Simulation results show that LMBER algorithm and AMBER algorithm are better than LMS algorithm in error rate performance, 1.5dB demodulation loss gap is observed especially in BPSK modulation system, and both AMBER and LMBER algorithms have a faster convergence speed than LMS algorithm. But LMBER algorithm has higher computational complexity than AMBER algorithm and LMS algorithm because of the estimation for noise power, which is not required in AMBER algorithm and LMS algorithm. Taking all of these characteristics of algorithms mentioned, we finally choose AMBER as our candidate algorithm.In order to solve the problem that burst signal has less amount of sequence, we use data reusing method to ensure the equalizer fully convergence, which means passing the training sequence and the information sequence received through the equalizer repeatedly. We also use real multipath Rayleigh fading channel model to test our algorithm, and the result shows that we realize the required goal of our project.Finally, we have implemented the design on Quartus ? software using Verilog HDL to verify the algorithm's timing sequence and function through Modelsim simulation, and we also use FPGA platform to prove the feasibility of the designing scheme.Although this paper only studies an equalization technique for short burst signals,but the research results can be also used in many other communication systems especially in military communication system and emergency communication system which have strict requirements for the bit error rate performance. This is all because of the obvious advantages MBER criterion has over MMSE criterion and the performance improvement caused by data reusing mothed.