EURASIP Journal on Wireless Communications and Networking

Table 1 Latency benchmarks of soft-output MIMO detection algorithms for a \(4 \times 4\), 16-QAM scenario

From: Joint high-dimensional soft bit estimation and quantization using deep learning

	EQ-Net	NN-Det	OAMP-	MMSE-	ZF-	SD
	(L)	(P)	Net2	SIC	SIC
Latency CPU [ms], \(B = 1\)	1.807	3.456	2.835	0.839	0.760	1.622
Latency CPU [ms], \(B = 16\)	2.256	3.953	3.227	1.109	1.012	19.52
Latency GPU [ms], \(B = 8192\)	9.107	17.251	30.501	8.248	6.253	–

B represents the batch size during parallelized inference. All values are expressed in milliseconds and represent average execution times for the entire batch

Back to article page