National Natural Science Foundation of China (No.61572114, No.61671390, No.61401273);National Postdoctoral Science Foundation of China (No.2015M570776);Key Program of Fundamental Research Funds for Southwest University (No.XDJK2016B002)
to satisfy the low-power dissipation requirement in mobile scenarios
a decoder with small memory size has attracted extensive attention.By decomposing the trellis diagram of the adopted turbo code
this paper proposes a memory reduced decoding architecture based on reverse recalculation.A modified Jacobian logarithm is specially investigated for the reverse recalculation
and the reverse recalculation in logarithmic domain and the realization structure are also presented.It shows that at the price of low redundant calculation complexity
the memory size is reduced by 50%
while the decoding performance is very close to that of the Log-MAP algorithm.The proposed decoding scheme is superior to other decoding architectures in terms of dummy computation complexity