MIMO 시스템을 위한 Lattice Reduction (LR) aided 검출 방법의 중요한 단점의 해결책으로 심별 판별의 신뢰도가 의심되는 경우 추가적인 보정을 해주면서, 계산양은 매우 작은 개선된 LR-aided 검출방법을 제안한다. LR 검출방법의 오류들은 주로, 마지막 lattice 역 변환 단계에서, lattice point들이 constellation point 바깥으로 벗어날 때 발생한다는 것을 발견하였다. 이를 기반으로, 제안된 기법에서는 LR-aided 검출방법을 통해 얻어진 lattice point들이 constellation 바깥쪽에 있는지를 확인하고, 바깥쪽에 있는 경우, 모든 constellation point들이 아닌, 얻어진 lattice point의 이웃하는 point들에 대해서만 ML(Maximum Likelihood) 검출을 수행한다. 제안된 기법을 사용함으로, 검출 성능은 ML에 근접하면서, 계산양은 아주 작은 효과를 기대할 수 있다.
An important drawback in Lattice Reduction (LR) aided detectors has been investigated. For the solution, an improved LR aided detection with ignorable complexity overhead is proposed for MIMO system, where the additional correction operation is performed for the case of unreliable symbol decision. We found that LR aided detection errors mainly occur when the lattice points after the inverse lattice transform in the final step fall outside the constellation point set. In the proposed scheme, we check whether or not the lattice point obtained through LR detection is out of constellation. Only for the case of out of constellation, we additionally perform ML search with reduced search region restricted to the neighboring points near to the obtained lattice points. Using this approach, we can effectively and significantly improve the detection performance with just a slight complexity overhead which is negligible compared to full searched ML scheme. Simulation results show that the proposed scheme achieves the detection performance near to that of the ML detection with a lower computational complexity.
