News & Analysis
WiMedia unveils strategic directions
Jack Shandle
12/7/2007 12:53 PM EST
Disappointing data transfer rates compared to the 480 Mbits/s PHY rate touted by Alliance members were tacitly acknowledged when Jeff Ravencraft, chairman of the USB-IF, reported that the "best-to-date" system-to-system data transfer rate recorded in laboratory certification testing is about 55 Mbit/s.
Ravencraft did not provide any information on the distance between the device and the host nor would he provide a low, average, or range of data transfer rates for all the certified and non certified devices tested. In the demos that followed the state-of-the-technology presentations, however, none of the devices were more than a meter apart—and, in most cases, they were within inches of each other.
The first independent, comprehensive, and publicly available test of commercially available UWB (both wireless USB and video) was recently been completed by octoScope, a RF engineering consultancy, and will be published by WirelessNetDesignline and its sister sites next week.
Culpability for the data-transfer performance gap was laid squarely on the fact that current Certified Wireless USB solutions use adapters for both the host PC and the peripheral device to complete the wireless link. The implementation is known as the HWA/DWA (Host Wire Adapter/Device Wire Adapter) implementation.
HWA/DWA implementations require that signals must traverse multiple protocol stacks. The WiMedia Alliance and USB-IF have accelerated programs that will eventually replace HWA/DWA implementations with "native" implementations in which devices know they are talking to Certified Wireless USB and can optimize for maximum performance and minimum power consumption.
Better performance can be achieved when a Wireless Host Controller Interface (WHCI) controller on the host talks to a native device. The WHCI specification describes the register-level interface for a host controller for wireless USB.
Ravencraft provided "best-to-date" lab data for three scenarios in addition to the previously mentioned HWA/DWA implementation.
- 70 Mbits/s for a host adapter to native device
- 160 Mbits/s for a native host to DWA device
- 220 Mbits/s for a native to native data transfer
Although there is unanimity of opinion within the WiMedia and USB-IF camps that device drivers are the root cause of current disappointing data rates, others in the RF community do not share that opinion. In any event, the compliance infrastructure for native implementations is being worked on now and solutions are expected sometime in 2008.
Over the past year, Certified Wireless USB demonstrations have frequently resulted in less than optimal host-to-device device link-up, or, association. This was still the case at yesterday's state-of-the-technology forum demos. Ravencraft said that in the future the Alliance and Forum would be adopting an association model based on Near-Field Communications (NFC).
Global spectrum availability has also been a cause for concern. The Alliance has defined six band groups in a range from 3 GHz to 10 GHz. Certified Wireless USB products announced to date are in Band Group 1 (3 GHz).
This presents a business problem because in this band group Europe and Asia want UWB products to implement an Detect and Avoid (DAA) protocol when they are in the presence of fixed broadband signals (such as WiMAX) or radar. But so far, there has been no resolution on what the DDA definition will be.
Some WiMedia members believe that the best route to market in these regions is to move to one of the higher band groups in the 6 GHz range where DAA is not an issue. But this raises a problem for OEMs of having different products for different parts of the world, and, the higher costs associated with that mode of operation. These members are hoping for a recalibration of preferred radio design targets up to 6 GHz to avoid the "multiple SKU" problem.
Stephen Wood, president of the WiMedia Alliance, said he anticipates that a DAA definition will be settled by the end of 2008. Although Ravencraft had earlier included "moving to upper band support" in his list of three priorities for the future, Wood would not acknowledge a recalibration of the technology to 6 GHz. Instead, products with multiple radios would be fielded, he said.
The program included several demonstrations from systems companies such as Belkin, I/O Gear and D-Link as well as drom silicon vendors such as Alereon, Intel, and WiQuest showing prototypes.
One of the most interesting—and the only one that operated at a distance of more than a few inches between the host and device—was Alereon's. That demo featured a 6-GHz radio (Band Group 6), three-inch external antennas, and native device drivers. It attained "top of the MAC" performance of 175 MBits/s. The chip set will be available in 2008, according to communications director Mike Krell.
Comments
grasor
12/8/2007 7:08 PM EST
Quoting Jeff Ravencraft and Stephen Wood in an article that appeared in the November 2006 issue of RF Design:
"Built from the ground up and optimized for a wireless medium, Certified Wireless USB supports high data-rate wireless connectivity and delivers the same speed (480 Mbps at three meters and 110 Mbps at 10 meters) and ease-of-use experienced with wired USB today. It is powered by the WiMedia UWB Common Radio Platform as developed by the WiMedia Alliance."
So... over the course of another year, the marketing hype and vapor-ware associated with WiMedia has once again proven to be mere PowerPoint presentations. In the January 2006 review of Belkin's wireless USB hub, the excuses were that the hardware implementing the device was pre-release, not fully WiMedia compliant. It measured less than 20 Mbps application layer throughput. Now, the latest and greatest stuff, after another year of work, comes in at a blazing 30-40 Mbps. I guess that will work for USB 1.1 at 12 Mbps, but will never get close to the 160 Mbps average real-world application throughput realized by wired USB 2.0 high speed.
Just wait until someone tests the WiMedia solution with 4 simultaneously operating piconets, one of the IEEE 802.15.3a requirements that will never be met by the current solution. The result will be MAD (Mutually Assured Destruction)! WiMedia needs a working DAA function to protect it's own devices from each other.
I don't want to rub it in too much, but what happened to the design wins that a Wi--- something company announced with Dell and Lenovo? I still haven't seen any manufacturer, sans one (Toshiba), with such a blind product staff that they would increase the BOM on their highly cost sensitive products by including another radio sub-system that performs worse than proprietary aftermarket 2.4 and 5.8 GHz solutions.
How long will the Venture Capitalists backing the WiMedia start-ups hold their breath? Even the Bluetooth SIG has given up on WiMedia's MB-OFDM solution.
In conclusion, it is highly unlikely that the WiMedia platform, in its current architectural state, will achieve its data throughput, range or power consumption goals. The current radio is too complex, and requires too many band-aids to merely operate in its sub 200 Mbps modes (thats raw PHY data rates, that you need to reduce by 20-30% for real world application data rates). Dont take my word for it, just hold off buying any WiMedia based products until one receives a Editors Choice award, or at least a review that show a reasonably operative communication system.
Sign in to Reply