Options
A Low-Complexity Interference-Aware Sphere Detector for Imperfect Channel Estimates
Publikationstyp
Journal Article
Date Issued
2016-07
Sprache
English
Author(s)
Institut
TORE-URI
Volume
65
Issue
7
Start Page
5337
End Page
5348
Article Number
7152970
Citation
IEEE Transactions on Vehicular Technology 7 (65): 7152970 (2016-07)
Publisher DOI
Scopus ID
In practical applications, an imperfect channel estimate over a noisy channel is inevitable. This paper investigates an interference-aware receiver in which colored-estimation-noise-based maximum-likelihood (CEN-ML) detection is employed to mitigate the impact of imperfect channel estimates. It is shown that the CEN-ML detection, which treats the channel estimation errors as colored noise, has superior performance and is robust to imperfect channel estimation. However, a CEN-ML interference-aware detector usually requires computational efforts that are too complex for multiple-input-multiple-output (MIMO) interference-limited systems. To reduce the complexity, this paper proposes new methods for both the preprocessing and tree-search stages of an interference-aware sphere detector (SD). First, a channel extended matrix is constructed using the information from channel estimation errors in a preprocessing stage. The channel extension model can reduce the complexity without performance loss. Second, two tree pruning (TP) criteria are developed in a tree-search stage to make the implementation of the CEN-ML SD feasible. The first criterion is based on a lower bound of the CEN-ML metric and can achieve the same result as the optimal brute-force method. The second criterion is derived from an approximated metric and can further reduce the complexity with small performance loss. Simulation results for typical Third-Generation Partnership Project Long-Term Evolution (LTE)/LTE-Advanced scenarios show that the new methods provide the ability to deal with the channel estimation errors and to achieve a better tradeoff between complexity and performance (e.g., in terms of throughput or error rate) in interference-limited scenarios. The proposed low-complexity CEN-ML SD can have a gain of more than 10 dB than the other investigated detection for 16-ary quadrature amplitude modulation (QAM) or 64-QAM at a target coded block error rate (BLER) of 10-2.
Subjects
Channel estimation
intercell interference
joint detection
Long-Term Evolution/Long-Term Evolution Advanced (LTE-Advanced)
network-assisted interference cancelation and suppression (NAICS)
sphere detector (SD)/decoder
Third-Generation Partnership Project (3GPP)