MIMO delivers range, beam forming delivers video

Next-generation wireless-LAN technologies extend 802.11g to deliver DVD-quality video.

By Maury Wright, Editor in Chief -- EDN, 1/19/2006

In the run-up to this year's rendition of the Consumer Electronics Show (CES), EDN decided to take a look at a couple of the latest wireless-LAN (WLAN) technologies that purport to carry video. By the time you read this, CES will have featured a spate of wireless-video demonstrations, because the consumer-electronics industry is seeking a "no new wires" way to move compressed digital-video streams, including HDTV streams, around a home. In addition to WLAN approaches, data-over-power-line technology and coaxial cable are challengers in this quest. But the WLAN arena is by far the most active, so EDN went into the Digital Den to see how some state-of-the-art WLAN products might perform in a multimedia-centric role. And there is improvement afoot.

As popular as WLAN technology—and more succinctly IEEE 802.11, aka Wi-Fi—is for data networking in the home and office, many believe it will proliferate even more once it starts carrying video in the home. However, even though several EDN staffers have been WLAN fans since the 1990s, our tests have consistently found that 802.11 (in all its alphabet soup of flavors) comes up short in terms of range and bandwidth. In fact, while we've applauded the convenience of Wi-Fi, we've also chronicled the fact that a single 802.11 access point (AP) struggles to cover a typical home, even for data networking. And for a network to carry audio or video streams, the gear must perform at a higher level because real-time multimedia carriage can't happen if the WLAN regularly drops, and has to retransmit, packets.

Lay of the LAN

Nine months or so I ago I converted my home WLAN to products from Belkin based on Airgo Networks' MIMO (multiple input, multiple output) technology. There's some ambiguity at this point as to exactly what qualifies as MIMO (see sidebar, "Spatial multiplexing and beam forming" and "The greed for speed," EDN, Feb 19, 2004, for details). Airgo pioneered a technology called spatial multiplexing, which is one form or element of MIMO, that essentially sends multiple data streams through the same RF channel simultaneously. The technology turns the normally villainous multipath into an asset, presumably delivering greater range and greater bandwidth at a given range.

I've tested a number or products over the years, seeking one that offered WLAN coverage throughout my modest 1500-square-foot house and backyard. Ironically, the 900-MHz Proxim Symphony product, operating at a measly 1-Mbps rate, handled the chore back in 1998, but Wi-Fi products failed consistently over the years. When 2.4-GHz 802.11b came along, it boosted rates, but range went down. Then 5-GHz 802.11a boosted rates further, with yet another reduction in range. Finally, 2.4-GHz 802.11g products, and proprietary technology extensions such as channel bonding, began to attack the range issue. Still, I had tested products with the latest chips from Broadcom, Atheros, and others, and never found one Wi-Fi AP that covered my entire property.

Click for larger image

The MIMO products, which Belkin initially branded as Pre-N, finally brought me full coverage. According to whom you believe, the "Pre-N" moniker either innocently promises MIMO technology today that's similar to what may be in the still-under-development 802.11n spec, or it misleads consumers into thinking that they are buying an upgradeable product.

Regardless, Belkin's products excelled where others hadn't. In earlier experiments, I had moved an AP into a fairly central location in my dining room rather than placing it in the logical place in my office near my cable modem. Yet I had still struggled at times to get a reliable signal on the patio outside the living room. And I had never witnessed a reliable signal in my son's bedroom in the back corner of the house. I haven't gotten around to measuring the data rate of the link with the Belkin equipment, but the WLAN utility reports "excellent signal strength" everywhere that I've used it. And I even tucked the Belkin router under my desk in my office.

New noise

Enter Ruckus Wireless (formerly called Video 54). A few months back, Ruckus briefed me on its new MIMO technology—in this case a beam-forming approach branded BeamFlex (see "Beam-forming WLAN implementation targets A/V usage," EDN, Sept 29, 2005). Ruckus doesn't build 802.11 chips, but rather sells a smart, multiantenna-element front end that dynamically sets antenna configuration and signal paths between sender and receiver to avoid interference. Ruckus also claims to augment the MAC (media access controller) layer with a software stack, called SmartCast, that improves multimedia streaming relative to what's in 802.11g and arguably even what's planned for 802.11n.

Ruckus sent me a pair of products: the MF2900 AP and MF2501 set-top adapter. The company probably won't sell this pair at retail, hoping a consumer-electronics company or service provider such as a cable company will do so. Ruckus estimates pricing of $169 for the AP and $129 for the client. The products are based on an Atheros chip set.

I decided to test the Ruckus products and also go back and fully evaluate the Belkin products. I connected the Ruckus AP to a 3-GHz Pentium 4 system that I located in my dining room, adjacent to where I'd always performed WLAN performance tests over the years. I connected the Ruckus client to my 1-GHz Pentium 3 notebook. Fairly quickly, I found that the Ruckus products covered my home just as well as the Belkin products, so I set about to measure data rates.

Click for larger image

Many software utilities and other methods exist to measure rates. I relied on a stopwatch and a Windows file transfer for my tests, as I have for all of the WLAN testing that I've done over the years.

With client and AP adjacent, I measured a maximum data rate of 17.8 Mbps. But I ran every test multiple times and the results varied, with one heat coming in at 11.9 Mbps. I suspect that the variation is due to interference over which I had no control. I do know that two other neighbors have WLANs than I can detect from my dining room.

I next went back to my son's bedroom. The distance is around 50 feet, although one never knows how the signal actually travels, through the walls or bouncing off them on its way down the hall? I was surprised to measure rates as high as 16.7 Mbps in that location, and never saw a rate as low as the minimum rate that I measured with the units adjacent.

Ruckus had suggested that I try to add noise to the equation by using a microwave oven. I placed one right next to the AP and set it to work boiling some water. I did not witness maximum or minimum performance with the microwave running, but I also could not discern for sure whether this was due to the microwave or simply the variation mentioned above.

Trying to further challenge the Ruckus technology, I took the client outside onto my patio, where I consistently measured rates around 11 Mbps. The range was about 60 to 70 feet, although the signal path may have been clearer than the one to the bedroom. Next, I took the client to the far side of my house on the property line (about 80 feet), where the signal almost surely passed through a stucco wall, and measured rates around 9.7 Mbps. The variation in multiple tests had become smaller as the data rate had dropped.

Video flights

To test video transmission, Ruckus supplied VLC Media Player, an open-source software package from VideoLan that can be used to stream video from one PC and decode and play on another. The company also sent a clip of the movie Monsters, Inc., stored in the DVD MPEG-2 format.

I tested streaming all around the house and yard, and never witnessed anything but perfect playback. The microwave had no adverse effect. Following a Ruckus suggestion—that I try a file transfer and a video stream at the same time—I found that video played undisturbed during the file transfer. Moreover, I measured surprisingly fast data transfers at the same time: 10.1 Mbps with the systems adjacent and 7.3 Mbps with the client outside my house on the property line.

Thoroughly impressed, I turned back to the Belkin products. I connected the router directly to the Pentium 3 desktop so that it was no longer connected to my home LAN. And I put the Belkin PC Card in my notebook. With the systems adjacent, I measured rates as fast as 18.8 Mbps, slightly better than the Ruckus performance. Again I witnessed random variation in repetitive tests, with rates dropping as low as 12.9 Mbps.

When I tested the Belkin products at greater distances, however, they did not perform to the Ruckus level. I measured rates in the 13-Mbps range in my son's room and the 8-Mbps range at my property line, still perfectly acceptable for data networking.

I also used the VLC software to see how the Belkin products stacked up in video streaming. But even with the systems adjacent to one another, I never witnessed perfect video on the client. The playback suffered from dropped frames and occasional blocking artifacts. Apparently there is something to the SmartCast software. The Airgo-based Belkin Pre-N products also include an optional QOS (quality of service) feature that is an extension to 802.11g for multimedia, but the Belkin products that I tested can't match the Ruckus on video right now. In fairness, however, a second-generation Airgo chip set is available in products already, and the company has also announced a third-generation IC.

Conclusions

I had hoped to test the Ruckus link with a high-definition video stream. Unfortunately, I couldn't locate HD content that the VLC software could handle before I ran out of time. When I get a chance to perform that test I will offer an update on our Web site.

For reference, I also tested the data transfer over a wired Ethernet link. With a 100-Mbps Ethernet switch between the two systems, I measured rates as fast as 79.6 Mbps. Clearly, WLANs still have some catching up to do. Belkin advertises its products as capable of 108-Mbps rates, while Ruckus highlights the 54-Mbps 802.11g spec.

For now, I'm sticking with the Belkin products for everyday use; the Ruckus client is a bit unwieldy because it's not in a PC Card form factor. By calling it a "set-top adapter," Ruckus clearly intends the device to be a living-room product, and in that setting, the form factor is fine. Beyond that, it will surely be easier for the company to integrate the technology's six antennal elements into a notebook lid than into a PC Card.

In 2000 I predicted that WLANs would eventually become the AV network of choice. Subsequently, performance issues with 802.11 forced me to flip-flop and instead pronounce that WLANs would never deliver the needed bandwidth over range. Today, I'm again ready to believe that 802.11n might handle whole-house video.

Spatial multiplexing and beam forming

No sooner had we heard the term MIMO (multiple input, multiple output) than it took on multiple meanings.

The term arose out of research at Stanford University, which Airgo Networks was first to commercialize. But as soon as MIMO became a hot term, purveyors of different types of technology latched on as well. Atheros was among the first to use the label on a product that simply had multiple antennas.

In any event, the ambiguous usage has now led to more descriptive technology labels, such as "spatial multiplexing" and "beam forming." Airgo is now using the brand True MIMO to separate the spatial-multiplexing technology that it offers from other technologies that are also being called MIMO.

In spatial multiplexing, multiple signal paths between transmitter and receiver exist simultaneously in the same frequency band, enabled by antenna diversity. This is the technology that will be mandated in the next-generation IEEE 802.11n specification now under development.

But don't dismiss other technologies, such as smart antennas or beam forming, as unimportant. Also called MIMO by some, beam forming can further extend bandwidth over range in spatial-multiplexed wireless LANs. But beam forming will likely be an implementation choice in 802.11n products, as opposed to a requirement.

MORE

Recent related articles:

Earlier experiments with WLANs: