Technology Rationale:
“Moore’s law” advances in programmable silicon integrated circuits have created an opportunity to develop intelligent or “cognitive” radios which can adapt to a wide variety of radio interference conditions and multiple protocol standards.  Such a cognitive radio would be capable of dynamic physical layer adaptation via scanning of available spectrum, selection from a wide range of operating frequencies (possibly non-contiguous), rapid adjustment of modulation waveforms and adaptive power control.  In addition, a suitably designed cognitive radio with a software-defined physical layer would be capable of collaborating with neighboring radios to ameliorate interference using higher-layer protocols. These higher layer coordination protocols could range from etiquette mechanisms all the way to fully collaborative multi-hop forwarding between radio nodes.  Such cognitive radios with both physical and network layer capabilities are expected to improve the prospects for both spectrum compatibility and interoperability between proliferating wireless data standards.

Technical Approach:
The network-centric cognitive radio architecture under consideration in this project is aimed at providing a high-performance platform for experimentation with various adaptive wireless network protocols ranging from simple etiquettes to more complex ad-hoc collaboration.  Particular emphasis has been placed on high performance in a networked environment where each node may be required to carry out high throughput packet forwarding functions between multiple physical layers.  Key design objectives for the cognitive radio platform include:

• multi-band operation, fast frequency scanning and agility;
• software-defined modem including waveforms such as DSSS/QPSK and OFDM operating at speeds up to 50Mbps;
• packet processor capable of ad-hoc packet routing with aggregate throughput ~100 Mbps;
• spectrum policy processor that implements etiquette protocols and algorithms for dynamic spectrum sharing.

The cognitive radio prototype’s architecture is based on four major elements: (1) MEMS-based tri-band agile RF front-end, (2) FPGA-based software defined radio (SDR); (3) FPGA-based packet processing engine; and (4) embedded CPU core for control and management.  These components will be integrated into a single prototype board which leverages an SDR implementation from Lucent Bell Labs as the starting point. 

Results To Date and Future Work Plan:
A prototype cognitive radio board is currently under development by the WINLAB, Bell Labs, GA Tech team.  The design of this board leverages the Lucent software defined radio platform shown in Fig 1, which is being upgraded to include a large FPGA (XCV4SX55) with configurable DSP slices for the modem and a second FPGA (4VFX100) for the network processor, as shown in Fig 2.   An MPC8560 processor serves as the CPU for execution of control and spectrum etiquette policies. 

At WINLAB, research work is currently in progress to investigate trade-offs between hardware and software solutions.  An optimal hardware/software partition is essential for a cognitive radio system which meets both performance and cost constraints.  The solution will include an example on how differing MAC protocols, such as TDMA and CSMA based protocols, could be implemented on the same system.

The digital processing board interfaces to a novel analog front end with an integrated tri-band antenna (shown in Fig 3) and a MEMS based VCO.  A proof-of-concept demonstration board is planned for the end of year 2 (Sept 2006), and several prototype platforms with full functionality are expected to be ready at the end of year 3 (Sept 2007).