Although long envisioned to dramatically improve spectrum utilization and to safely share multiple frequency bands without interfering with licensed users, building and deploying energy efficient cognitive radios with dynamic spectrum access have proven to be a challenging task. This project proposes an energy-efficient cognitive radio system that addresses its many practical challenges including a low-power wideband radio that operates in multiple frequency bands and a means to seamlessly and efficiently coordinate the operating bands between communicating unlicensed users. The proposed work presents a complete integrated solution, ranging from radio implementation, signal processing algorithm development, and network connectivity. These different layers of abstraction have been jointly optimized to enable a low-power and robust cognitive radio system.
The proposed energy-efficient cognitive radio (CR) system, which is jointly optimized at the circuit, signal processing, and network layers, operates as follows: 1) A secondary user (SU) employs a novel compressed sampling receiver based on polyphase sequences implemented via mixer harmonics to sense concurrently the entire frequency span of interest. 2) Since the communicating band between SUs is unknown, data is transmitted by sending the same signal at low power levels across the entire frequency span of interest while nulling energy in the primary user (PU) bands. A novel sub-Nyquist transmitter, which can be viewed as a dual of the compressed sensing receiver, is proposed for this purpose. 3) The receiving SU uses the same compressed sampling receiver for sensing PUs to receive the SU signals. This is achieved by projecting the received signals to PU's null space, whose frequency locations were estimated during sensing. By appropriately selecting the basis functions set in the analog front-end, most of the received energy can be added coherently. 4) As SU energy is spread across the entire frequency span of interest, the resulting channel impulse response (CIR) is richly scattered. The CIR itself, therefore, can be used as users' signatures for achieving multiple-access communications. Furthermore, by leveraging on techniques from multiuser detection theory, high spectral efficiency can be achieved.
