You are not logged in.
Mango has just posted the first release of the 802.11 Reference Design. Please visit http://warpproject.org/802.11 for the latest design files and documentation.
Offline
Hi,
On the 802.11 Reference Design page it says "The 802.11 Reference Design is designed for Mango WARP v3 hardware. Previous generations of WARP hardware are not supported, as the design requires tools and FPGA features which preclude porting to Virtex-II Pro or Virtex-4 FPGAs." Can you explain what "tools and FPGA features" are utilized that does not allow this design to run on WARP v2?
Thanks
Offline
The primary one is AXI-based interconnect, which Xilinx introduced a few ISE releases ago but did not port to Virtex-4. All cores in the XPS design (the ones we designed and the standard Xilinx cores) use AXI bus interfaces. The PHY design itself also relies on a few Sysgen cores which are not supported on V4 (FFT 8.0, FIR Compiler 6.3, floating-point arithmetic in the equalizer).
Offline
Ok, so the AXI interconnect is actually a hardware change that was not included in the Virtex-4 series.
1. Since this is a hardware limitation of WARPv2 I'm guessing it is extremely unlikely to see this reference design working on v2?
2. I think I read somewhere that the WARPv3 RF interfaces support 40MHz channels (hopefully I'm not making that up). Is that the maximum bandwidth? Is there any hope for 80 or 160MHz with this hardware?
3. Throughout the 802.11 Reference Design guide it is mentioned how several features are currently under development (such as 16-QAM, 64-QAM, antenna diversity, MIMO, etc.). At this time would you be able to divulge which features will most likely be release earlier than others, or is it too early to tell?
4. I am interested in 11n research so I would be preferable to support for the FMC-RF-2X245, allowing up to 4 spatial streams. Do you see this happening anytime soon?
Thanks again
Offline
1. Yeah, without Xilinx support for AXI on V4, porting this reference design to WARP v2 would be very difficult (back-porting all the cores from AXI to PLB, possibly redesigning the PHY pipelines to meet the resource/timing constraints of the V4FX100, etc.).
2. Each WARP v3 interface supports 40 MHz of bandwidth (the maximum bandwidth of the MAX2829 transceiver). This is also true for each interface on the FMC-RF-2X245 module. Expanding to 80 or 160MHz may be possible through channel bonding (tuning multiple interfaces to neighboring channels and treating the multiple sets of I/Q like a single wide channel). However, driving 160MHz of BW from a design is a challenge in itself. It wouldn't be as easy as changing a parameter in the 802.11 Reference Design to take advantage of all of that bandwidth. In short, I think this possible, but it would be an engineering effort beyond changing a parameter in the 802.11 Reference Design.
3. Additional rates (16-QAM and 64-QAM) are working in simulation and should be added to one of the next releases. Switching diversity is also straightforward, so that would probably be shortly after adding in the additional rates.
4. We definitely intend to add MIMO modes in the future, but this is (at least) many months off. We need to study the MIMO modes introduced in 11n and 11ac to figure out which subset is both feasible and useful.
If specific modes would be useful in your project, please let us know (we can't promise anything, but any feedback on what's useful to researchers is helpful).
The next few releases will focus on finishing 11a/g support and robustifying the whole design. We also plan to add better hooks for running experiments with many nodes.
Offline