Performance of efficiently-encodable iteratively-decodable block codes

Principles for designing efficiently-encodable (EE) iteratively-decodable block codes are investigated. We assumed encoding is performed by an erasure-filling decoder and are interested in minimizing encoding time, subject to maintaining good error performance in decoding. We focus on a design parameter, called the recovery vector n, which controls the encoding speed of EE codes. When designing powerful block codes, this vector affords great flexibility in setting the encoding speed. Both analytic and simulation results indicate that faster encoding speed eventually implies a degraded error floor performance, and in practice the tradeoff between encoding speed and error floor performance should be evaluated to meet the actual application. In addition, an extra design rule is proposed for EE low-density parity-check (LDPC) codes, and we show that the popular irregular repeat accumulate (IRA) codes and efficiently-encodable rate-compatible (EERC) codes can be obtained by using such rule.