Abstract:
Collaborative decoding is a scheme that achieves receive diversity
by exchanging decoding information among a cluster of physically
separated receiving nodes. In our previous work, we proposed a
technique in which information about the most reliable bits is
exchanged. This most-reliable-bit exchange scheme provides
performance close to the conventional maximal-ratio combining
technique, while providing significant savings on the amount of
information that must be exchange. In this work, we approximately
evaluate the error performance of the collaborative decoding
procedure when nonrecursive convolutional codes are used. The
analysis is based on the assumption that the extrinsic information
generated in the collaborating decoding process for nonrecursive
convolutional codes can be approximately described by a Gaussian
distribution. A density evolution model based on a single MAP
decoder is used to obtain the statistical characteristics of the
extrinsic information. We use this approximation to evaluate the
error performance of the MAP decoders in the collaborative
decoding process.