Multi-sensor integration on the NCUBE hypercube computer

Abstract

This paper presents the quantitative performance results obtained from converting a large sequential Multi-Sensor Integration (MSI) program into a concurrent form, and executing the program on an NCUBE hypercube computer for several test case scenarios. Each sensor track was assigned to a different node of the hypercube where computationally intensive calculations were performed, A 4-node hypercube was found to be the optimal size for the test case scenario discussed here, where speed-up factors of 1.5 to 1.7 were obtained. A completely concurrent MSI algorithm is also presented.