NB-LDPC matrices and simulation results

Several NB-LDPC matrices are proposed in order to evaluate and compare the performance of NB-LDPC decoder. Matrices are described using the 'alist' format . The 'alist' format is modified for NB matrices using the q-ary 'alist' format .

The matrix element are given as the power of the primitive element. GF(64) constructed using the irreductive polynomial Q[X] = X6 + X + 1. Similarly, GF(256) is defined thanks to the irreducible polynomial Q[X] = X8 + X4 + X3 + X2 + 1. The binary representation of the finite field can be obtained using the following matlab function build_GF.m

NB-LDPC matrices on GF(64)

NB-LDPC matrices on GF(64)
Nbin Kbin Rate GF Nsymb Ksymb dv dc Girth #g* #g+2 #g+4 dmin alist Author Simulation results
96 48 1/2 64 16 8 2 4 8 36 0 0 11 alist1 [KL] [ML] 4 Bubbles
576 288 1/2 64 96 48 2 4 8 4 24 20 ? alist1 [KL] 4 Bubbles
576 480 0.83 64 96 80 2 12 6 256 2136 16128 6 alist1 [KL] [ML] 4 Bubbles 8 Bubbles min-max
864 720 0.83 64 144 120 2 12 ? ? ? ? ? alist1 [UBS] 4 Bubbles
2304 1152 1/2 64 384 192 2 4 16 1164 58560 ? alist1 [KL] 4 Bubbles
9216 8064 0.83 64 1536 1344 2 16 ? ? ? ? ? alist1 [UBS] 4 Bubbles

NB-LDPC matrices on GF(256)

NB-LDPC matrices on GF(256)
Nbin Kbin Rate GF Nsymb Ksymb dv dc Girth*, #g, #g+2, #g+4 dmin alist Author Simulation results
96 48 1/2 256 12 6 2 4 6, 8, 15, 24 ? alist1 [KL] 4 Bubbles
128 64 1/2 256 16 8 2 4 8, 36, 0, 0 13 alist1 [KL] [ML] 4 Bubbles
128 64 1/2 256 16 8 2 4 8, 36, 0, 0 14 alist1 [U-NBPB] 4 Bubbles 8 Bubbles
256 128 1/2 256 32 16 2 4 ? ? alist1 [U-NBPB] 8 Bubbles
512 256 1/2 256 64 32 2 4 ? ? alist1 [U-NBPB] 8 Bubbles
512 256 1/2 256 64 32 2 4 8, 16, 0, 0 ? alist1 [KL] 4 Bubbles
576 480 0.83 256 72 60 2 12 4, 18, 255, 1809 ? alist1 [KL] 4 Bubbles 8 Bubbles min-max
1152 960 0.83 64 144 120 2 12 ? ? ? ? ? alist1 [UBS] 4 Bubbles

------