MAXIMUM SHARED RESOURCE ARCHITECTURE FOR RECURSION COMPUTATION IN TURBO DECODERS

Abstract

Long Term Evaluation (LTE) has been used to achieve peak data rates in wireless communication
system. Turbo codes are used as the channel encoding scheme. MAP algorithm has been used as a decoding
scheme. Complexity in MAP algorithm is reduced by implementing the algorithm in log domain giving rise
to Log MAP algorithm. The main objective of this work is to reduce the complexity of state metric
computation by employing of various algorithms and these algorithms differ only by their implementation of
correction terms. The ACS unit is implemented using constant log MAP algorithm, linear log MAP
algorithm, MAX log MAP algorithm, multi step log MAP algorithm and hybrid log MAP algorithm. The
state metric calculation is implemented with the help of radix-4 Add -Compare-Select (ACS) unit. The
distance calculation involved between two concurrent computations of state metric can be shared among
them which give rise to Maximum Shared Resource (MSR) architecture. The proposed implementation of
these algorithms leads to reduction in the power dissipation, propagation delay and the number of logical
elements used for the recursion computation in turbo decoders used in LTE system. The MSR architecture
for recursion computation reduces the number of LUTs by 12.1% when compared with the existing.

Author

T.Akila, R. Vijayabhasker