Mark Twain famously said (or almost said) “rumours of my death have been greatly exaggerated”. Just when you thought it was safe to stay at home and live with slow speed wireless data transfer, UWB has performed a similar reincarnation, appearing to rise, Lazarus-like, from its grave with the announcement of two new chipsets from Samsung and CSR.
UWB has had a chequered history of ups and downs. Last year, when I started writing my book “Essentials of Short Range Wireless”, I planned a chapter on it as it seemed to be experiencing something of a renaissance. Half way through writing the book, a number of the key chip companies folded and I removed the chapter. It looks as if I may have acted prematurely.
Why the resurgence of interest? UWB has had a turbulent history, with many of the start-up companies supporting it going bust as the industry embarked on its love affair with ever faster variants of Wi-Fi. The answer comes back to the classic divide between the PC and mobile phone industries and the feature that separates them more than anything else: one has a power cord and the other doesn’t.
That may sound like a glib comment, but it exercises many of the best design minds within each industry. Ask most phone users what they want most in a phone and it’s one feature – the battery life, which manifests itself as talk time. Most smartphone users have wistful, fond memories of a simpler phone (typically a Nokia 3310) which had a battery that only needed charging one per week. Phone manufacturers and resellers will tell you that poor battery performance is still the one thing that is most likely to cause a customer to return a handset.
In contrast, poor battery life has long been seen as something a laptop user needs to live with. The industry’s approach is generally that bas long as the next laptop has more memory, a bigger display, faster Wi-Fi and a recordable Blu-Ray drive, who cares? They’ll argue that it comes with a power cord, and spends most of its working life attached to it.
Both products suffer from the same technological rat race, with competitors vying to add new features. But the fundamental difference is that phone owners expect their phone to work for at least a day without being charged, whereas very few laptop owners would dare take their laptop out of the house in the morning without a power adaptor.
That’s why UWB is interesting, particularly for the mobile phone. When GSM phones first came out the few models that could support a data connection had a maximum speed of 2.4kbps, which was all the networks were capable of. Today networks can provide over 1Mbps and phones can store and transfer GBytes of data. The problem is that it takes energy to transfer data wirelessly. As the volume of data increases, the amount of energy needed to transfer a few GBytes exceeds the capacity of a normal phone battery. So if we’re planning to stream video from a phone, the industry needs to develop a technology that can do it efficiently.
The requirement to stream video has upped the stakes, requiring a throughput of 35Mbps or more. That’s beyond the capability of 802.11g, which is why the industry is moving to 802.11n. 802.11n achieves its higher throughput by splitting the signal into multiple streams, using multiple antennas, transmitters and receivers. To work well, antennas need to be separated by several centimetres, which is fine for a PC, but a major problem for a mobile phone. That limits mobiles phone that use 802.11n to using a single stream; a configuration known as 1×1. With a single stream, at the most complex QAM64 coding, 802.11n can still deliver a theoretical maximum throughput of 65Mbps, and a practical one of around 40Mbps, but at a cost. The power needed to support this rate using 802.11n is high. Sufficiently high that it can reduce the battery life of a phone to less than half and ruin the talk time. Moreover, with only one channel, the connection becomes directional. As you move around with the phone, the throughput will drop significantly. If you’re using your phone to give a business presentation, streaming to a projector, that turns 802.11n from a compelling solution to a disaster.
To address this, standards designers need to look at the efficiency of wireless data transfer. In the race for higher speed, that’s generally been a point of largely academic interest. As a very rough rule, as the throughput of a standard goes up, the energy per bit needed to send data over the air goes down as more information is packed into each bit. However, this improvement tails off as the underlying protocol starts to become the dominant factor, limiting performance. Also, as data coding get more complex, the transmitter and receivers needs to work harder, reducing any power savings. Which means that at some point you generally need to develop a new standard if you want to get to the next level of performance.
That’s where UWB comes in. By being designed for very high throughputs as a starting point, it doesn’t have the baggage of a low speed heritage that bedevils the 802.11 specifications. At throughputs of around 30Mbps, data sheets show that typical 802.11n transmitters consume around 350mW, whereas UWB transmitters are less than a tenth of that. Moreover, UWB was designed for streaming using an isochronous MAC, whereas 802.11n has to live with the inconvenience of a contention based access, which wastes more power. For a phone, that means video streaming changes from a poor possibility to a real application. Which is what I suspect is driving the Samsung and CSR announcements, as both have a strong interest and history in supporting innovation in mobile phones.
Both companies’ chips appear to be aimed at the mobile market, either for phones, or for battery powered, handheld devices. Samsung’s solution is a complete two ship set – the S3C2680 and S5M8311, which provides a complete wireless USB compliant solution, small enough to be implemented in an SD card and targeted at portable devices.
There’s less information on CSR’s solution, which was “outed” in Incisor – the wireless industry’s main newsletter. They’d picked up the information from the Bluetooth SIG’s qualification site. This puts information about all Bluetooth products and components in the public domain and revealed that CSR has qualified a version of their Synergy stack, which supports the Bluetooth 3.0+HS High Speed standard, but which uses a new Ultracore UWB chip that is WiMedia certified, operating in the 6-9GHz band. There’s a brief technical document describing it that can be downloaded from the Bluetoooth SIG site.
The presence of these won’t stop the Wi-Fi bandwagon trying to get 802.11n into handsets. Atheros, along with two more specialist chip companies – Redpine Signals and Nanoradio, are working hard to optimise their 802.11n 1×1 solutions to bring the power consumption down. However, physics dictates that they’re not going to get to the levels that are possible with UWB.
And the PC industry is already looking to move on to 60GHz, with a new standard emerging from the Wireless Gigabit Alliance (WiGiG). This follow the same PC route of assuming a power cord and a mains socket, transmitting at powers that would make your phone battery weep, so isn’t really competition for mobile devices.
Will UWB make a comeback? It’s still too soon to tell. It will be the best part of a year before we see either of these chips in a commercially available phone, but I’d expect to see some interesting demos before that, maybe at CES or MWC. If these capture the imagination of user, then it may really be a case that UWB could eventually be the preferred wireless standard for high speed data on the mobile handset.