WARP Project Forums - Wireless Open-Access Research Platform

The PHY models were developed in simulation and tested primarily in real-time in hardware. Sysgen 8.2 and 9.1 (but not 9.2, for some infuriating reason) are fast enough that the full Tx/Rx of multiple packets can be simulated in a few minutes. This is sufficient to verify the operation of the signal processing and control logic. The trickier part is hooking the PHY up to the rest of a full wireless system. The interplay of AGC, packet detection, radio control and the PPC/OPB/PLB can't be readily simulated. We debug the full system in hardware, making extensive use of ChipScope, digital I/O and analog I/O to observe the operation of the complete system.

The OFDM reference design is SISO by default. The OFDM model in the design supports both SISO and spatially multiplexed 2x2 MIMO modes, and can switch between them on-the-fly (it's a runtime software setting, not a hardware parameter).

There would be some resource savings with a strictly SISO model, but it wouldn't be anything near half. Most of the signal processing blocks are time-shared between the MIMO streams. The receiver's single FFT core, for example, processes data from both antennas serially.

Our PHY was designed for non-mobile setups, so the coherence bandwidth assumed for 64-subcarriers in 10 or 20MHz bandwidth is a good fit. The model could be extended to more subcarriers. The bulk of the processing chain would work as-is (it's one long pipeline), but a lot of the control logic would need tweaking.