Generalized Error Locating Codes with Soft Decoding of Inner Codes

Abstract:
In this paper we propose to consider a generalized error-locating code (GEL-code) as a possible candidate for data transmission systems that require high code rates along with strict requirements on wrong decoding probability. The paper describes the construction of the GEL-code and the algorithms for encoding and decoding. The main idea is to implement a soft decoding of the inner codes. It will be shown that such approach results in better performance than hard decision decoding of inner codes. We present a method for constructing GEL code (with the minimal redundancy) that guarantees that the probability of wrong decoding will be less than required one (for a given channel error probability). Numerical results for a maximal achievable code rate depending on input signal-noise ratio on code symbol are presented.