Your browser is not supported. Please update it.

6 March 2015

RIoT @ MWC: The networks roundup

•    Sense of Sigfox feud as LoRa Alliance formally launches
•    Wirepas claims 700,000 node mesh network, breaks 1m deployed barrier
•    Sequans begins march to Cat 0 LTE with Cat 1 modem and Verizon trial
•    Spirent dives into IoT network testing, warns of legacy complications

Sigfox wins and LoRa launch lay grounds for quiet but bitter rivalry

At MWC, Sigfox was on show with a booth full of devices that were using the low-power radio protocol, as well as announcing the launch of the Sigfox network in Portugal. Amid a dozen or so devices, we saw a HydroKo water valve, a Gaslog gas meter and the Z#BRE box that tracks care-worker visits to a home.

In a hail of questions, we learned that the Sigfox deployment inside the Spanish Securitas alarm systems was primarily chosen as a means of avoiding cellular jamming devices – which are not hard to come by these days. We also enquired what the recently secured $115m funding would be spent on, and were told that a national US network rollout was in the planning stages.

The Project Mustang work is also a little further off than we anticipated, with Sigfox’s product marketing manager Eric Nicolas telling us that Mustang has a 3-year duration to build a proof of concept. The project would see Sigfox terrestrial networks linked via satellites, and is being conducted in partnership with aerospace firm Airbus Defense and Space, the French research institute CEA-Leti and French engineering business Sysmeca.

We also tagged along to the official launch of the LoRa Alliance, after the group announced plans to form a working body to develop the low-power wide area network (LPWAN). We won’t name names, but during a few probing questions about a potential LoRa-Sigfox rivalry, there appears to be a significant difference of opinion on the viability of Sigfox – based largely on Sigfox’s current lack of bi-directional communication, according to several attendees at the LoRa event, who it seems missed Sigfox’s quiet upgrade of its network infrastructure to support bidirectional communication. We can’t take the high ground there, as we missed the news too – with Sigfox later telling us that it prefers to under-report and over-deliver, rather than promise the world and miss its targets.

The LoRa speakers, from Semtech, IBM, KPN, Bouygues Telecom, and Actility, stressed that LoRa’s benefits arise from 4 main characteristics: a cloud-based management system; its close alignment with IEEE 802.15.4; minimal energy requirements; and smooth versioning to eliminate useless overheads in the codebase.

After the increasingly raucous crowd finished drowning out the speakers, and an iPad was handed out as an attendance prize (thankfully to someone who was at least paying attention), RIoT went looking for hands to shake – and learned about the potential rift between LoRa proponents and Sigfox’s.

We heard praise for Sigfox’s marketing department, but a good dose of scorn for its technology. We’ll see if this ever spills over into public, but it seems that not all LPWANs are created equal. RIoT will be keeping tabs on this potential feud.

Wirepas builds largest mesh network with 700,000 nodes

Having recently broken a world record for the largest concurrent mesh network, with a fairly staggering 700,000 nodes on the same network, Finland’s Wirepas was at MWC to tell the world that it had now successfully connected its first million devices in its Nordic deployment – mostly comprised of smart meters from Aidon.

Using a technology called Pino, Wirepas’ network doesn’t feature any central gateways or routing hardware. Instead, every device in the network is a node that will self-regulate and optimize its connections to neighboring devices – which has created a mesh network some hundred times bigger than the next largest existing network, according to CEO Teppo Hemiä.

The onboard software stack is tasked with regulating the device’s network connections, at a scale that make the likes of ZigBee and Z-Wave seem rather inadequate – given their infamous instability issues that crop up in mesh networks that hit the hundreds of devices, which can be traced through frustrated forum posts FAQs.

Wirepas PINO is a software stack that live on the device. Its software stack is provided for a one-off royalty fee, and the design is currently completely hardware-agnostic, meaning it can be deployed on any number of MCUs and RF units in almost all hardware designs.

It operates in the 868MHz and 915MHz ISM bands, the unlicensed but still managed RF bands. This means that while it has to conform to some rules that regulate maximum transmission power, the frequency is free to use. This is the tradeoff that must be made in the unlicensed band, as the licensed spectrum is prohibitively expensive for new entrants due to its popularity among the cellular operators.

But the increased chance of interference, particularly in congested areas, shouldn’t be too problematic. Mesh networks rely on having multiple paths to relay traffic – so even if a device is locally interfering with the Pino installation, the message should still be able to route around the problem. If the network is completely jammed up, the devices have the intelligence to wait a period of time before trying again. In theory this means that mesh networks are reliable and resistant.

And if Wirepas’ claims are true, then its deployed base is proof of this line of thinking. With internet-connected gateways deployed through the network, its autonomous ease-of-use should make this an attractive proposition to developers looking for a very specific criteria – of low-power, highly resilient and low-cost devices.

Wirepas makes the point that centralized network controllers represent single points of failure, and that the distributed autonomous approach is better for the types of installations that Pino has so far been deployed in. Hemiä also noted that ZigBee’s contention-based MAC layer is known to fall over at scale.

Sequans begins march to Cat 0 LTE with Cat 1 modem and Verizon trial

Sequans felt like one of those companies that you could talk to for hours, discussing the ins and outs of its industry and consequently running late to the next appointment. Nonetheless, we managed to fit a lengthy discussion of the role of LTE in the IoT markets during our MWC visit, and learned that a national US trial of LTE Category 1 had just been completed in partnership with Ericsson and Verizon – using Sequans’ single-mode LTE chipsets.

LTE Category 1 (Cat 1), otherwise known as the 8th release of the LTE spec by the 3GPP standards body, defines a much lower bandwidth version of LTE that can potentially be used for the types of M2M application that the current 2G/GSM technology is currently prevalent in the industry. The 3GPP has begun looking at the lower-speed specification recently, after an initial push to grab headlines with massive speeds.

The current push to move to LTE from 2G (now) and 3G (eventually) is being driven by the mobile network operators’ (MNO) desire to reuse their existing 2G spectrum and re-farm it into LTE-ready deployments. With the benefits of LTE in IP packetization, reduced latency, and IPv6 addressability; if the MNOs eventually sunset their 2G infrastructure, the existing and potential future M2M and IoT deployments need to find new homes in LTE – hence Cat 1 and eventually Cat 0.

However, Cat 1 and eventually Cat 0 require new hardware at the infrastructure level to function, and this was the main thrust of the Sequans-Ericsson-Verizon trial – which aimed to prove that the new Cat 1 technology was interoperable with existing LTE deployments, so that smart meters or remote sensors don’t brick someone’s Netflix stream.

All three companies are calling the trial a success, with the lower data rate Cat 1 tech (up to a maximum of 10Mbps download and 5Mbps uplink) proving itself capable. As Cat 1 applications, such as remote sensing, don’t require a lot of bandwidth (preferring a longer battery life and smaller form factor to blazing-fast upload speeds), Cat 1 has been designed to lower the device cost and power consumption – at the expense of the (unnecessary) bandwidth.

Speaking to Craig Miller, Sequans’ VP of Worldwide Marketing, we discussed whether Cat 1 was in a position to surpass 2G for remote sensing, and were told pretty resolutely that this would not be the case. Cat 1 LTE is suitable for a few apps, but for the majority, it doesn’t clearly provide an advantage over the established technology.

However, over the course of our MWC appointment, Miller explained that the wider LTE industry had surged ahead with the potential top-speeds of LTE in Cat 4 and up at the expense of the lower end efficiency. Consequently, it took until Release 12 for the 3GPP to directly address M2M applications, and eventually define Cat 0 – a rather important addition to the LTE standard that will actually provide clear benefits over the 2G and 3G standards for those that transition to the new technology.

At the heart of it, Cat 0 promises to provide a 1Mbps download and upload, using only one physical antenna, compared to the two that are required for other LTE implementation. Consequently, this promises to reduce the cost of the device modules, as well as reduce the complexity of the software that is needed to manage them, and the hardware on which to run that – thanks to the parallel reductions in the size of the frequency channels that the modules use. As such, a Cat 1 chip isn’t really software upgradeable to Cat 0, as it still packs the redundant (and expensive) LTE hardware – offsetting any savings in bandwidth usage or power consumption. Cat 0 requires a ground-up development approach.

So for operators who make the transition to LTE Cat 1 and 0, which requires significant capex and R&D investment in their networks, the opportunity to provide connectivity services for a new class of IoT device awaits. But ‘waits’ is the key word here, as Cat 0 rollouts are not going to appear until 2016 or early 2017 according to Miller.

On the module side, once the correct cellular networks exist, Miller said that the opportunity exists for Cat 1 and Cat 0 to outperform the legacy 2G devices in terms of module cost. Cat 0 is aiming to outperform the 2G module pricing, by focusing solely on Cat 0. Adding a multimode radio to a module, able to receive 2G, 3G and 4G or any combo) adds a lot to the BOM cost, according to Miller.

We will have our ears to the ground for updates on Cat 1 and Cat 0, but given the expected timeframes, it seems that the MNO side of things will lag behind the rival M2M technologies that are surging ahead in the space – in particular the sub-GHz protocols such as Sigfox and LoRa.

Spirent dives into IoT network testing, warns of legacy complications

Spirent makes its money by testing networks, typically cellular and IP-based ones, and so it has unsurprisingly leapt into the IoT markets, as well as branching out into exploring the user experience on a network and not just the benchmark results.

Saul Einbinder, Spirent’s VP Venture Development, told us that the IoT was presenting new challenges at both the device level and in the wider network. He added that the arrival of M2M has been akin to open-sourcing the network, as it has brought thousands of devices into cellular networks that were never designed to accommodate them.

Instead of bandwidth-hungry handsets, M2M devices typically use orders of magnitude less data than a smartphone. On the surface, this may sound like a non-problem, however, Einbinder explained that every time one of these devices tries to connect to a network it requires a series of messages from the network before it can begin to send its message.

As Ross Cassan, the director of marketing for Spirent’s fixed and mobile networks and applications business, for a twenty minute YouTube session or a five minute phone call, this process is not such an issue. The networks and the protocols that keep them ticking over were built to accommodate that use case. However, with the M2M devices, the same process of allocation is required for a message that will last a few seconds at most. In addition, M2M deployments are set to out-number handsets in a market, meaning that a poorly managed network is in danger of falling over it can’t handle the requests.

This is where Spirent hopes to make its money, by making and leasing new testing equipment that emulates the strains that an M2M/IoT deployment could place on an MNO’s network equipment – whether it’s 2G, 3G, LTE, WiFi, Ethernet, BLE or 802.15.4.

With more affordable testing units, offered to developers on a lease-basis, the types of startup developers and new entrants typically associated with the IoT could have access to enterprise level testing – that doesn’t requires the extensive knowledge that a networking engineer might bring to the table, which the connected toaster developer might be lacking.

By way of the Elevate M2M Testing Framework and the Landslide network emulation software, Spirent hopes to allow developers to responsible build devices that aren’t capable of causing nationwide network outages – and avoiding the complications that arose out of a 2013 Android update that tried to conserve battery by aggressively disconnecting from the network, which in turn doubled or tripled the reconnection traffic when the handsets began trying to join again.

When we put the question of LTE Cat 1 and Cat 0 adoption for M2M applications to Cassan and Einbinder, the pair pointed to the EU’s eCall (emergency calling) standard for use by vehicles that might be involved in traffic accidents. eCall has now been delayed many times, and has been bumped all the way down to a 2018 release and deployment in 2020.

The issue with eCall is that it mandates a 2G connection, and as we heard from Sequans, there is the potential that 2G spectrum will have been re-farmed for LTE by this point in time – leading to the slightly absurd situation that vehicles due to be built in 2019 will be shipping with legacy connectivity, instead of shiny new LTE.

In this case, of legacy devices from the future, Spirent’s reps point to the spanners that can be thrown into the works when slow moving standards agencies, in this case government-backed standards, essentially require MNOs to keep 2G networks online to cater a niche that might be better served with new technologies. After all, 2G first appeared back in 1991 and is pretty archaic by today’s standards.

But this is all good news for Spirent – as it wouldn’t have any business if the networks all worked perfectly. So whether it’s government meddling or rogue toaster developers, it looks like there will always be a need for companies like Spirent who will make sure that the networks that underpin the looming hyper-connected world continue to function.