TY - GEN
T1 - On the performance of single LDGM codes for iterative data fusion over the multiple access channel
AU - Del Ser, Javier
AU - Garcia-Frias, Javier
AU - Crespo, Pedro M.
AU - Manjarres, Diana
AU - Olabarrieta, Ignacio
PY - 2010
Y1 - 2010
N2 - One of the applications of wireless sensor networks currently undergoing active research focuses on the scenario where the information generated by a data source S is simultaneously sensed by N nodes and therefrom transmitted to a common receiver. Based on the received information from such N nodes, such receiver infers the original information from S potentially more accurately than in the case of a single sensor. Often referred to as the CEO (Central Estimating Officer) problem [1], in this scenario we propose the use of single Low Density Generator Matrix (LDGM) codes for the transmission of the information registered by the nodes over the Multiple Access Channel (MAC). The corresponding receiver iterates between a soft demodulator, the set of N LDGM decoders and a soft-information fusion stage. Simulation results for the AWGN MAC channel show that 1) the proposed coding scheme outperforms the suboptimum limit assuming separated Slepian-Wolf distributed coding and capacity-approaching codes; and 2) the end-to-end Bit Error Rate (BER) performance is lower bounded, for increasing N, by the error floor due to the inherent ambiguity of the MAC channel when dealing with correlated sources. This paves the way for future research aimed at applying concatenated coding schemes to this setup.
AB - One of the applications of wireless sensor networks currently undergoing active research focuses on the scenario where the information generated by a data source S is simultaneously sensed by N nodes and therefrom transmitted to a common receiver. Based on the received information from such N nodes, such receiver infers the original information from S potentially more accurately than in the case of a single sensor. Often referred to as the CEO (Central Estimating Officer) problem [1], in this scenario we propose the use of single Low Density Generator Matrix (LDGM) codes for the transmission of the information registered by the nodes over the Multiple Access Channel (MAC). The corresponding receiver iterates between a soft demodulator, the set of N LDGM decoders and a soft-information fusion stage. Simulation results for the AWGN MAC channel show that 1) the proposed coding scheme outperforms the suboptimum limit assuming separated Slepian-Wolf distributed coding and capacity-approaching codes; and 2) the end-to-end Bit Error Rate (BER) performance is lower bounded, for increasing N, by the error floor due to the inherent ambiguity of the MAC channel when dealing with correlated sources. This paves the way for future research aimed at applying concatenated coding schemes to this setup.
KW - data fusion
KW - LDGM codes
KW - multiple access channel
UR - http://www.scopus.com/inward/record.url?scp=78049294696&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-15428-7_5
DO - 10.1007/978-3-642-15428-7_5
M3 - Conference contribution
AN - SCOPUS:78049294696
SN - 3642154271
SN - 9783642154270
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 47
EP - 57
BT - Multiple Access Communications - Third International Workshop, MACOM 2010, Proceedings
T2 - 3rd International Workshop on Multiple Access Communications, MACOM 2010
Y2 - 13 September 2010 through 14 September 2010
ER -