Qualcomm and Broadcom may have made it up in the courtroom, but Qualcomm has fired some pretty powerful shots directly at Broadcom in the marketplace ‘ trying to usurp its role as the global leader in Wi-Fi chips ‘ attacking in-home video transmission and low power Wi-Fi for handsets.
If Qualcomm was to have the type of effect that it has had in the HSDPA markets, where from a standing start it took more than 50% market share ‘ simply by being first to release each HSDPA speed landmark first ‘ it could potentially decimate Broadcom’s future Wi-Fi revenue base.
Both products were on show at the Computex show in Taipei and at Connections 2009 in Santa Clara, the first being a 4 by 4 MIMO wireless chip with multiple in multiple out antenna.
This is the first time that Qualcomm has made a run at the digital home market ‘ a business which has been in turmoil for the past decade ‘ with every other chip business on the planet shooting at the problem of how to move HD video streams around a home.
There are wired systems such as MoCA, HPNA and HDMI, all good for different parts of the equation ‘ and all about to be potentially superseded by newer cable technologies and faster and faster wireless systems.
Most Ultra Wide Band technologies have either failed or been discredited by association, and 60GHz wireless appears to be on the remote horizon, including WiGIG, just announced, which leaves us with Wi-Fi. The big problems with Wi-Fi is that traditional Wi-Fi drops about 15% of its packets and unless you have a return path to request packet resends, and special software, as Ruckus Wireless does, to chart and re-direct traffic on channel which have better performance, then it’s tough to reliably support video. Ruckus products are significantly more expensive than traditional Wi-Fi.
One approach, which Qualcomm is now endorsing, has been to push towards MIMO and in 2007 Apple was one of the first to deliver a product built around the pioneering work of Atheros, which has many patents in this area. The Apple Airport Extreme used Atheros chips which preceded the 802.11n standard, and included patent pending Signal-Sustain technology. It operated in dual-concurrent 2.4GHz and 5GHz spectrum. These chips featured 2×2, 2×3 or 3×3 MIMO smart antenna arrays and Atheros claimed even then that the top of the range products would hit 600Mbps and offer real world video throughput up to 180 Mbps. Today Atheros has still to reach those heights.
However Atheros has long been seen to lead this quiet little niche of Wi-Fi, and it has openly said that it makes little sense to go beyond 3 x 3 antenna arrays. The new Qualcomm product uses 4 x 4. Now Qualcomm has been in the Wi-Fi business for a while, not with discrete Wi-Fi chips, but by adding Wi-Fi to chips that do other things ‘ part of a SoC chip, so it’s not a total newcomer here. But the 4 x 4 move is definitely new and out there on the technical edge where Qualcomm is most comfortable. The only other vendor to sample a 4 x 4 802.11n product has been start-up Quantenna, which has not gone commercial yet, but claims its product can be configured in a pair, to deliver up to 1Gbps.
The problem is that each transmit or receive antenna in a MIMO set up requires a lot of circuitry to be placed on a chip – power amplifiers, RF and IF filters and mixers, a baseband filter, digital to analog converters and vice versa for the return journey. So the gain you get in terms of sending more signals is often outweighed by the loss you get in paying for the extra circuitry and a step level increase in complexity. Four by four means multiply all this logic up 16 times over.
Qualcomm’s N-Stream WCN1320 chip, claims to have solved all these problems and created a throughput of around 600 Mbps. Qualcomm says it is designed to distribute multiple simultaneous HDTV streams throughout the home.
‘Qualcomm’s advanced 4 x 4 MIMO technology sets a new performance milestone for home media streaming devices,’ said Mike Concannon, senior vice president of Connectivity and Wireless Modules for Qualcomm CDMA Technologies. ‘Our N-Stream solution delivers a new level of performance and range, enabling next-generation whole-house wireless LAN coverage.’
Of course 600 Mbps can only compete with compressed formats for video ‘ a typical compressed 1080p HD stream might require 8Mbps these day, though it is a pretty moveable feast and 18 Mbps is not unheard of. So moving two or three of these around should be child’s play for an advanced MIMO 802.11n system. However 2 or 3 Gbps is what you’d need for the same video uncompressed and many efforts in this market are aiming at decompressing video at the home gateway and distributing uncompressed (mostly wired) video traffic around the home.
The WiGig specification which should emerge by late 2009 is a blueprint for building Wi-Fi-like chips which can transmit 1.5 Gbps but since it uses 60 GHz, will not be able to get through a single wall. So this is being targeted as transmitting across a room, not around an entire home. And even that is tough and may not emerge in anything like its roadmap timescale.
We have recently discussed two other of these technologies – both wired, one out of China called DiiVA, which supports 13.5Gbps of bandwidth dedicated to video plus a hybrid bidirectional channel and another out of Israel called Valens Semiconductor, which has 20 Gbps on its cable roadmap using Ethernet protocols over coax. But most of these technologies have trouble with whole home distances (Valens claims it can deal with this) and the entire world seems to be looking to Wi-Fi to solve this problem ‘ at least for now and 600 Mbps Wi-Fi might just do the trick.
Qualcomm says it has developed sophisticated algorithms which take advantage of the multiple transmitters and receivers to increase data throughput, extend range and overcome interference ‘ but the company as yet has given out no technical details about how this is possible, except to say that the chip is built in a 65-nanometer CMOS process and is 12×12 mm in size and combines an embedded applications processor, media-access controller, digital baseband, radio-frequency transceiver and system power-management. That’s a BIG chip and much could go wrong with this and yields and heat dissipation will both likely be problems ‘ at least initially.
NXP said it has been working with Qualcomm and this chip on set tops, while Applied Micro has worked with it on access points, network attached storage and home gateways. Qualcomm will need many more friends than that to dominate this sector.
The key to this chip is that it is ready to ship in samples now (second quarter) and it’s part of a new range which also takes in low power 802.11n in the WCN1312 chip for handsets and other mobile devices. This mobile Wi-Fi chip supports data rates up to 72 Mbps all aimed to improve the experience at a handset used for watching streaming video or surfing the Internet.
What we would like to think is that Qualcomm is dropping us a large hint here that the two chips may well have a future together, which is something that we have always said. But this doesn’t quite hold together we don’t think.
A handset wants to reach out to all of the media resources in a home, and if Wi-Fi connected to a server, then the speed of connection is vital if files are going to copied for viewing on the way to work or away from the home. If a video file is an hour long in SD it might be roughly 1 GB in size, but in QVGA that would come down to maybe 200 MB. Now that might copy in around 25 seconds at 72 Mbps. That is if all of the bandwidth on the handset could be harnessed, and if this was not a battery drain, what with all that Flash memory writing going on.
We think it is more likely at these speeds that a micro USB cable would be used so the Wi-Fi chip is just to get a slightly better streaming or HTML experience on handsets ‘ nothing more. We have always pictured the 1 Gbps promised by UWB as being the type of connection for copying files to handsets and taking them with you. In fact to use the handset as a carry case it is more likely that you will be taking SD or even HD files with you, and then the copy time at 72 Mbps is way too long to copy last night’s soccer onto your phone to view on the train.
So this really is just Qualcomm looking to dominate a market for a chip that has been now been proven ‘ chips for dual band handsets. Previously this was a territory purely for Broadcom ‘ take over a chip market as it enters the volume phase. In the past Qualcomm has tended to create brand new chip markets with brand new technologies. And it just so happens that among the major suppliers of low power Wi-Fi chips to handsets are Atheros and Broadcom which has just finished pursuing Qualcomm so wholeheartedly through the courts.
At least the WCN1312 comes in a more sensible 7×7 mm package, is built in low-power 65nm CMOS and the company says it is optimized to work with other Qualcomm chips. It probably needs to be smaller still in future iterations. The WCN1312 chip is scheduled for sampling in the second quarter, and expected to be in volume by the fourth quarter.
Atheros is also launching more of its XSPAN branded MIMO chips for the WLAN market at Computex this week taking the 802.11n design up to 300 Mbps, but if the performance claims of Qualcomm are to be believed, it has leapfrogged the Atheros performance in one go with its own MIMO efforts.
Atheros clearly is not rich enough to sustain the type of legal campaign that Broadcom has undergone with Qualcomm, but it may feel slightly aggrieved that it has made all the running and filed many of the patents in MIMO, especially in the way it relates to Wi-Fi. But Qualcomm could always point to the fact that Atheros has never launched its own 4 x 4 antenna array, so perhaps we will see Qualcomm annex the home video Wi-Fi market without another legal row, though clearly one is possible.
In the set top market Broadcom will find that it has many friends, but if customers and colleagues count for much in this business, Qualcomm will certainly make huge dents in the handset Wi-Fi chip business.