Your browser is not supported. Please update it.

24 July 2015

Neul’s legacy: three Weightless specs and Huawei’s ‘4.5G’

In 2011, a UK start-up called Neul proposed a low power wireless protocol to run in the briefly fashionable TV white spaces spectrum. This would target long range applications from rural broadband to smart metering. Neul had made the wrong spectrum choice and was hedging its bets on the target market, so the auspices were not good, and indeed, the company struggled to gain market presence, and was acquired by Huawei almost a year ago.

However, the company’s legacy may prove far more important than that short history implies since it has, indirectly at least, given rise to two candidates to be the leading LPWA (low power wide area) network technology in the emerging smart city market. Neul itself is an important element of Huawei’s Cellular IoT development, which it is already pushing into commercial implementations under the ‘4.5G’ banner and which the Chinese firm is also proposing as a 3GPP standard for inclusion in LTE Release 13.

And Weightless, the name adopted for the original Neul technology as it bid to be a de facto standard, now has two sets of specifications published. These are not as Neul envisaged them – as interest in the white spaces waned, the Weightless SIG was forced to make a rapid transition into the ISM band (868 MHz in Europe), where most licence-exempt LPWA technologies – such as Sigfox and LoRa – live. This resulted in Weightless-N, whose silicon supplier (and major technology contributor) is now Nwave instead of Neul. And this week, the SIG made up for time lost in the spectrum shift, by announcing a second protocol, this one with bidirectional support, Weightless-P.

Weightless-P is a sub-GHz protocol that aims to provide the bidirectional functionality that isn’t supported or prioritized by Weightless-N. Like Weightless-N, which is almost entirely based on technology contributed by Nwave, Weightless-P is mostly derived from the portfolio of M2Communication (M2Comm), a Taiwanese business whose Platanus protocol has a very similar list of capabilities to those of the new would-be standard.

M2Comm likened its contribution to Weightless as “Platanus on steroids,” with some similar aspects but also “some tweaks and new features added to achieve reliable, bidirectional 2km range in a cluttered urban environment”.

Weightless-P still has to go through the usual channels and review processes of its SIG, but Weightless SIG CEO William Webb is confident that it will be ratified and approved by the end of the year. The group is also open to contributing it to other standards bodies, if that would be advantageous. The first Weightless-P hardware is expected in early 2016.

In an interview with Rethink’s RIoT publication ahead of the announcement, Webb said that P was filling a gap in the range of standards the SIG offers – which still includes Weightless-W for the white spaces, though this has effectively been abandoned following the slow progress associated with global spectrum allocation and harmonization.

At the chip level. Webb said that the PHY layer in P differs from that of both N and W, and uses a more classical radio approach, with 12.5KHz channels rather than the 200Hz UNB (ultra narrow band) approach of N. Current SoCs should be suitable for the new protocol.

Webb said that, although P requires a bit more silicon and stack software compared to N, both chips should have similar bill of materials cost at the same volumes. But of course, it is the volume of chip production which is the main factor in cost (hence why the relatively complex WiFi chips can be built at such low cost), and so far, no LPWA technology has achieved mass volumes.

Importantly on the economics front, the SIG is aiming to ensure that all Weightless devices share the same application layer interfaces, so that developers have an easier time of it. In addition to reducing complexity, this shared application layer also means that joint P + N networks could be built, perhaps with P devices acting as gateways for the more unidirectional traffic found in Weightless-N edge devices – an interesting proposition for developers or clients looking to deploy large scale hybrid use networks.

Webb added that Weightless was aiming to cater for 90% of the possible applications in the market, without pigeonholing itself into a particular niche. Aiming for general usage, Weightless is leveraging the fact that MNOs have realized they face a technology gap, at least in the short term, where LTE does not support the ultra-low power levels required for some low data rate, wide area appplications such as smart lighting, parking and metering. Cellular advocates will point to LTE Cat 1 and Cat 0, but we have only just seen the first Cat 1 devices to make it to market, from Sequans, with Cat 0 and LTE-MTC further off.

There are various routes by which the cellular community aims to seize back the high ground in smart cities. LTE-MTC is an adaptation of existing specifications, which will be incorporated in Releases 12 and 13, although some companies are already creating stepping stones based on the existing but previously ignored LTE Category 1 specs. In Release 12, which was frozen in March, Category 0 is defined, with maximum throughput of 1Mbps and optional half-duplex operation.

Nokia claims it will pip its rivals to the post with the first LTE-MTC field trial, to be conducted with KT of Korea, in the fourth quarter of this year. Huawei is also active in LTE-MTC, and claims its pre-standard ‘4.5G’ technologies will easily make the transition to full standards. It is working on a smart parking pilot, in partnership with China Unicom Shanghai and China Unicom Research Institute, and on other LTE-MTC projects with Deutsche Telekom and Middle Eastern carriers.

There are also no fewer than six ‘clean slate’ proposals for review by the 3GPP (see inset), which would start almost from scratch in creating an IoT-capable standard compatible with LTE and LTE-MTC, and with full convergence at the Release 14 stage. Which of these variants will be approved to proceed to be a work item for standardization is not clear but the lucky winners will be processed in parallel with LTE-MTC, also for Release 13. And it is notable that the three most prominent have significant input from the LPWA community (the other three candidates, from Ericsson, Nokia and Qualcomm, appear to have attracted little support so far, which raises interesting issues about the shifting balance of power in the cellular world once it moves to IoT).

Among these is Huawei’s Cellular IoT, which initially appears to harness Neul LPWA technology, but in cellular bands, and provides a convergence path to LTE Release 14. Vodafone has come out in support of Cellular IoT, which is likely to be commercialized next year as well as becoming part of the 3GPP standards. An interesting outcome – considered quite possible by some stakeholders – is for this effort to converge with that of the LoRa Alliance, which is also seeking to push its technology into 3GPP Release 13 via a licensed-band implementation in GSM bands.

The rising number of LPWA contenders:

Weightless-N has now been deployed in London and two Danish cities (Copenhagen and Esbjerg), admittedly at very small scale, but it is early days for the new protocol. Nwave has more extensive deployments of the technology that preceded Weightless-N, and M2Comm also has established deployments of its Platanus protocol – which shares a lot of characteristics with Weightless-P.

Weightless-P will offer fully acknowledged uplink and downlink, using a narrowband modulation scheme in the sub-GHz ISM bands (typically 868MHz, 915MHz, and 433MHz) but also some of the more unusual bands like 470MHz and 780MHz). Idle state power consumption is listed as less than 100uW, with GMSK and offset-QPSK modulation.

Forward error correction (FEC), Automatic Retransmission Request (ARQ) and Adaptive Channel Coding (ACC) are listed as prominent features, and the protocol will support roaming for mobile applications. A 2km urban environment range is claimed, with the obvious caveat of longer range in more open or rural settings.

The FDMA+TDMA narrowband channels, 12.5KHz wide, support channel hopping to counter interference or noise, and aim to provide an adaptive data rate (ADR) ranging from 200bps to 100kbps at a maximal transmit power of 17dBm – allowing the devices to turn down the data rate to increase range, or increase the rate to send messages in a shorter time span, which can improve the free capacity of a network by allowing more devices to speak over a given period. Time-synchronized base stations for scheduling transmissions are also a feature that should improve spectral efficiency.

Adaptive data rates are one of the key features of Semtech’s LoRa protocol, and in this sense, Weightless-P is pretty similar. Sigfox also offers a form of ADR, but with a lower upper threshold – and Sigfox and Weightless-N are the more comparable protocols, favoring more (but not always totally) unidirectional traffic, flowing from the edge to the center.

Another feature common to both Weightless-P and LoRa is the use of Spread Spectrum, although Semtech’s modulation is listed as proprietary in its data sheets, while the nature of Weightless-P as an open standard plus open hardware should appease more vocal or staunch supporters of open source platforms.

As for the political side of things: Sigfox is a proprietary protocol available on a wide range of third party hardware, with a per-device subscription paid to Sigfox; LoRa positions itself as an open standard, but to use it requires using Semtech silicon, or chips from a supplier that will be paying a royalty or license to Semtech; and membership of the non-profit Weightless SIG entails a fee/royalty too – admittedly only £900 p.a.

Nothing is entirely free in the IoT, or any industry for that matter, but Weightless P and N look more open than the leading two LPWAN technologies. Weightless posits that it provides for multivendor support and ensures interoperability between manufacturers, so that its users have a choice of silicon and devices – with royalty free IP.

Derrick Wei, the CEO of M2Comm said “Weightless-P will breathe new life into an IoT market that is waiting on a reliable LPWAN technology. Many IoT sectors require reliability to be addressed with ultra-low energy wireless communications not accessible through operator services today. The slow adoption of wireless IoT solutions highlights a huge technology void that Weightless-P will address without requiring custom hardware. Our current technology deployments in the retail and industrial sectors are a powerful indication of the future success we anticipate for Weightless-P.”

The 3GPP’s LPWA efforts:

Within the 3GPP GERAN framework there is a study item (expected to be completed by August 2015), which should feed into Release 13 standards. Called ‘Cellular System Support for Ultra Low Complexity and Low Throughput Internet of Things’, it would reuse legacy network RAN and core infrastructure.

Six key variants are under consideration, two based on evolved GSM/GPRS, the others on a ‘clean slate’ approach:

  • Ericsson-backed ‘Extended-Coverage GSM (EC-GSM)’: Based on legacy GPRS there are concerns that this solution cannot achieve the 160dB link budget using its message repeat approach.
  • Nokia-backed ‘Narrowband Hybrid Modulation’: Also based on legacy GPRS, this solution uses stronger coding to meet the link budget target, but only one contribution made so far.
  • Vodafone/Huawei/Neul ‘s ‘Clean Slate’, which Huawei labels ‘4.5G and officially labelled ‘Narrowband M2M’’ Based upon the Weightless LPWA technology from Neul, this uses subsets of channel bandwidths of either 180 kHz or 200 kHz (15kHz DL and 5kHz UL) to support a link budget of 160dB, for high density deployment.
  • Qualcomm-backed ‘Narrowband LTE’: This splits a 200 kHz channel into 2.25 kHz narrowband ‘tones’ but wasn’t strongly backed at the most recent GERAN meeting.
  • SIGFOX’s ‘Cooperative Ultra Narrowband’: This is understood to be the SIGFOX/Telensa system, which is also being pushed through ETSI.
  • Semtech/LoRa’s ‘Combined Narrowband with Spread Spectrum for Cellular IoT’.