Cooperative Spectrum Sensing in the Vehicular Environment: An Experimental Evaluation

We present the results of TV band sensing experiments performed in a suburban environment in Iizuka, Japan. This work adds to our previous work in which we proposed to exploit dynamic spectrum access techniques to facilitate use of the white space to solve shortage of spectrum for vehicular networks. Here, we focus on the terrestrial TV band spectrum sensing, which in addition to the "whitespace database", is one of the methods envisioned to identify vacant channels. A convoy of four vehicles senses multiple TV channels using energy detection. Local signal strength measurements collected from the USRP software defined radios (SDRs) are combined at the trailing vehicle which acts as the fusion center. We post- processed collected measurements to improve accuracy of local sensing outcomes by removing temporal drift and differences between the sensors remaining after calibration. We tested three conventional fusion algorithms: 1) selection combining; 2) simple averaging; and 3) weighted averaging. The results show that collaboration provides a significant improvement in the sensing performance. We observed minor differences in performance among used fusion algorithms.