Reduced Initial Searching Radius for Sphere Decoder

Sphere decoding (SD) has been proposed as an efficient algorithm in digital communications. Its complexity is polynomial for both an infinite and finite lattice which makes the real-time implementation of a maximum-likelihood (ML) decoder feasible. The choice of an initial radius for the SD has a great impact on the trade-off between the performance and complexity. In this paper, we present a new initial radius selection scheme for the SD based on the constellation modulation type and the target bit error rate (BER). The proposed scheme accurately define an initial radius of the SD, which reduces the radius redundancy and ensures that at least one lattice point is inside a sphere. Complexity analysis shows that the proposed scheme has much lower computational complexity than that of the conventional one with zero decoding failure rates meanwhile ensures no performance loss.