Research on RFID collision detection algorithm based on the underdetermined blind separation

Aiming at the collision problem of RFID tag communication, we proposed a new method for identifying the RFID recognition of the blind separation of source signals, which is based on the label collision and recovery. In the case that the number of tags is more than that of the reader, we established a RFID anti collision model based on the blind separation. In the same time considering the non full sparsity of RFID signal, we used an improved two step extraction method. Firstly We estimated the base vector of part of the source signal by the sample of the unit interval of signal. Then we extracted the corresponding source signal with the minimal interference so that the necessary conditions for the recognition of the mixed matrix can be avoided. After that, We have carried out a simulation comparison in terms of throughput. The result showed that the method of this paper is better than the conventional blind separation algorithm and ALOHA algorithm in terms of throughput and it is more suitable for the collision identification and recovery of RFID non fully sparse signal.