Incremental redundancy: A comparison of a sphere-packing analysis and convolutional codes

Theoretical analysis has long indicated that feedback improves the error exponent but not the capacity of memoryless Gaussian channels. Chen et al. [1] demonstrated that a modified incremental redundancy scheme can use noiseless feedback to help short convolutional codes deliver the bit-error-rate performance of a long blocklength turbo code, but with much lower latency. This paper presents a code-independent analysis based on sphere-packing that approximates the throughput-vs.-latency achievable region possible with feedback and incremental redundancy for a specified AWGN SNR. Simulation results indicate that tail-biting convolutional codes employing feedback and incremental redundancy perform close to the sphere-packing approximation until the throughput reaches the limit of the system's ability to approach the channel capacity.