The 'Weightless' proposal could enable ICT systems to better manage the huge shoals of data that will stream through the airwaves once the Internet of Things scales up.
Last year saw initial proposals for a new wireless technology standard for machine-to-machine (M2M) communications. M2M, also known as the Internet of Things (IoT), enables remote devices such as sensors and meters to send data into a central network. Called Weightless, this emerging standard is optimised for machine communications, and predicated on delivering device costs of below $2, battery life of up to 10 years, and ubiquitous operating coverage across a country. It makes use of a concept in radio spectrum access known as 'white space', interworking with TV transmissions in the UHF band.
A number of milestones have since been achieved, including: the fabrication of the first Weightless chip; recent announcements by Ofcom to pilot white space applications this year; and the publication in April of the Weightless standard specification in a final stable 'version 1.0' format. Completion of the standard is of course a key step in the development of this nascent technology; so how have we got to this point in the space of 12 months – and what have been the challenges and hurdles involved in bringing a brand-new technology toward market?
First, some backstory. Weightless was conceived by the Cambridge technology company Neul after it perceived a lack of a long-range machine communications technology able to deliver on the emerging vision of M2M or IoT. With all early work on Weightless conducted by a single technology company, one option would have been for it to remain a proprietary technology; however, experience teaches that there are no successful proprietary wireless technologies – all of the wireless systems routinely used are open standards. Weightless also needed to be 'open' to deliver on its vision.
Neul's initial work to show that a technical solution for this market was possible had resulted in a rudimentary specification that was considered around 60 per cent complete. Perhaps the most difficult decision was whether to use an existing standards body (ETSI or the IEEE, say) or to establish a new standards entity. With Neul's experience of Bluetooth, which had created its own body, there was some bias toward that route.
Another determinant was speed: Neul was funded by venture capital; the time to deliver working equipment conforming to an open standard was limited. So the company established a standalone body called the Weightless SIG (special interest group). It is possible that the Weightless SIG might move Weightless into an existing body once it's completed ' and indeed it is working with ETSI to assess whether this is appropriate.
Setting up the Weightless SIG was an arduous process. The first step was to recruit the 'launch Promoters' – credible companies that would form the initial board and show, at launch, that the standard had widespread support. These companies could not be found through marketing as, by definition, this was pre-launch; they had to be recruited through direct personal contacts, meetings, and detailed discussion. This stage was onerous, with key companies opting in and out through the process (often as individual representatives were reassigned internally), and legal teams working on initial memorandum-of-understanding documents that would form a basis for subsequent legal work. This stage took about eight months, at the end of which ARM, CSR, C&WW, and Neul itself, were onboard.
There then followed a year-long process to agree the legal framework and intellectual property rights policy for the SIG. That so much time (and expense) was needed might seem hard to credit; but every legal concern had to be addressed and circulated to all board members, who typically responded with subsequent concerns that then had to be addressed and resolved. However, the SIG had put in place an 'Early Adopters Agreement', a temporary legal structure that allowed work on the standard to commence before all the formal legal documentation was in place. This meant that much of the development work could proceed before all the legal details were fully agreed.
With the board in place, members needed to be recruited in sufficient numbers to populate the various sub-groups and provide enough effort to draft the standard. This was achieved through a launch event run by wireless community association Cambridge Wireless in September 2011. Its participation was a major asset as it could market the event and manage the logistics. The first 50 members of the SIG were recruited from the 100+ attendees to this event. From February 2012 full work started on the standard, with six sub-groups, each meeting weekly to develop their area. Progress was rapid. By the end of 2012 the standards work was almost complete, and it was time to grow the membership and the credibility more strongly. Having a concept for a technology is one thing; having working hardware is more challenging; and having production-standard systems another major step change. Neul went through several stages. The first base stations and terminal solutions – designed for internal evaluation – were made in small volumes of fewer than 10, and were 'cost no issue' builds using powerful FPGA (field-programmable gate array) devices and ample memory. This was because multiple changes to the software were anticipated as the devices were tested and the standard matured, and Neul did not want to be constrained by, for instance, insufficient memory in the terminal. At the same time that Neul started building these it also started on the chip design process.
This was a big risk, with so little of the specification defined, and with little experience of trialling the various aspects of the standard, such as the underlying wireless sub-system; but the lead times on chip developments are such that Neul had to take a best guess at the chip parameters, build extra memory on to the chip as contingency, and be ready to modify the design as it learnt more from testing.
Before the first hardware had been built Neul was asked to supply systems for trials such as the Cambridge White Spaces Consortium programme, a trial designed to help Ofcom translate its proposals for licence-exempt access to white-space spectrum into a secure enabling framework. This entailed rapidly hand-building more equipment to deploy six base stations and six terminals around Cambridge. These units were prototypes, and needed frequent software updates as the trial progressed. Despite being put to a purpose they were never intended for they performed well: indeed, the base stations are still in place and operating.
With a clear demand for tens or hundreds of early-stage devices from many companies interested in trialling Weightless – or even deploying small-scale commercial solutions – it became clear that a new version of the hardware was needed that was cost-reduced, production quality, and could be fully supported with remote monitoring and 'over-the-air upgrade' capabilities. Based on what had been learnt from the prototype Neul came up with a new design that used fewer of the expensive FPGA chips, had a better-designed radio system with increased performance, and was encased in a solid box for outdoor operation.
Neul contracted an external production company to build the product in volumes of hundreds, and then went through the process of evaluating initial samples, enhancing the design, getting the first batch and evaluating the failure cases, further enhancing the design and then transferring the build and test to an external company: a process fraught with issues from external housing redesigns that Neul's manufacturer had applied unilaterally because it thought it would improve things (it had the opposite effect) through leads that could not quite bend enough, to pole-mounting solutions that had bolts too long for the poles encountered in practice. All solvable ' but irritating. Neul has also nearly finished the design of a development kit that can be used by those building applications to check that their software and system design works correctly in advance of implementation using the custom chip.
The chip development was on-going apace using an array of simulation tools to check and hone the design. In the fourth quarter of 2012 the schematics were sent to a Taiwanese fabrication plant, and in November Neul received back the first batch of 200 chips. Testing was tense – if a serious design error had occurred the resulting cost and delay of the rework would have been a very major blow. Happily the chip – codenamed 'Iceni' – worked better than had been hoped.
Some of these 200 chips are now being shared with partners who are building modules and trialling them within their terminal devices. Neul has just submitted its 're-spin' of the chip with design modifications to fix the few issues that had been noted. It will receive back many thousands of this version in the third quarter of 2013 and, if it is content with the performance, may request larger volumes for initial Weightless solutions. The company also hopes that other semiconductor companies – such as CSR – will be bringing its own Weightless chips to market in the near future.
Alongside this, but less obviously visible, has been a massive software effort to build the operations and maintenance centre (OMC) that enables monitoring of all terminals, remote diagnostics of performance issues, remote software updating, provision of network performance statistics and much more. Indeed, Neul's software team is larger than its hardware and chip design teams combined.
Weightless has broadly been positively received by those in the ICT industry and elsewhere who recognise its potential. "Without low-cost communications between devices, the so-called Internet of Things is likely to remain only wishful thinking," commented Pike Research senior analyst Bob Lockhart. "The anticipated growth in M2M communications, as more 'smart' devices find their way into homes and industry, argues for low-cost approaches to [this kind of device-to-device] communication."
Not everyone has given Weightless an unequivocal thumbs-up. Some critics point out that cellular technology already provides good M2M coverage, and is indeed penetrating the M2M market, making a newer technology unnecessary. It's an obvious point to make, but the Weightless SIG's view is that it overlooks a crucial consideration: while cellular can address a small part of the market that can tolerate relatively high device costs and has external power, there is perhaps 90 per cent of the predicted market of 50 billion devices for which cellular is not suited; and this is the market Weightless aims at.
The Weightless SIG is a diverse community: predicting what it might develop, and by when, is inexact. Best estimates are that we will see widespread silicon availability in autumn 2013, network deployment in smart cities in the UK, US, or Singapore by the end of 2013, modules in the thousands for applications like smart metering by spring 2014 – with national network rollouts during mid-2014. It's feasible that the first million Weightless devices could be in place early in 2015, growing towards the first billion by 2017/18.
Prof William Webb, FREng, FIET, FIEEE, is Weightless SIG CEO,a director of Neul (www.neul.com), forthcoming IET deputy president, and author of 'Understanding Weightless' (Cambridge University Press).