Comparison of clustering algorithms for data detection in Multispectral Camera Communication

In this work, the performance of an optical camera communication (OCC) system is compared using several clus-tering algorithms in a cluster-based data detection approach. Furthermore, a multispectral camera is utilized to capture the thermally induced spectral variations in light-emitting diodes (LEDs). Thus, more than one channel can be attained from the same device. The results of this paper prove that using a clustering method can enhance the bit error rate (BER). Finally, different training sets were used to fit the clustering models underlining the impact on the system performance.