Wirepas Wirepas Blog Wide Area Mesh – Too good to be true?

September 27, 2017

Wide Area Mesh – Too good to be true?

By Wirepas

There is a saying – One gets what one orders. Or something similar - you get what you measure. Sounds like a good approach and straight forward to get what you need and want. Or is it?

It works well when your need is something you are certain about and quite short term or when you set one clear parameter target and build a plan to achieve it.

Building a complex system and maintaining it is however a much more complex process. Unfortunately, the needs and KPIs are often contradictory. Focusing on one parameter at the cost of others may even take you in to a dead end as certain early design choices may prevent to achieve other KPIs with the same design, system approach or chosen business model.

On the other hand, from time to time there are paradigm shifts that disrupt the market and when you look back they are obvious, although that was not the case in the beginning. Famous example is the transistor of which role and value as a digital switch was not clear when invented, but is today seen as one of the most important inventions in the history of technology.

In wireless connectivity, the digital cellular systems have been in incremental evolution already for more than thirty years and are still going strong. If we simplify, the main driver since voice has been data within a power budget. In IoT connectivity the drivers in the beginning were wide coverage, low power and low end-device cost.

The coverage has been built with two main approaches. LPWAN (Low Power Wide Area Networks) technologies focus on as long range as possible by taking the radio bit rate very low and by using as high output power as feasible. Another approach has been mesh technologies by building a protocol that enables multi-hopping (PAN, Personal Area Networks). Both approaches have some pros and cons and are used for different purposes. LPWAN is good for a wide area, low bandwidth, and sparse installations (apart from deep indoors and basements) while mesh networks have been used in relatively small and local networks, therefore mainly in homes. Both lack higher density and node amount support due to lack of collision avoidance or good spectrum efficiency. Also, both requires preinstalled infrastructure (base station, concentrator, or dedicated routers).

These two approaches have fulfilled relatively large portion of the IoT use cases and verticals until now. However, going forward, it is expected that the requirements will diversify beyond range and power. Frost and Sullivan released a white paper about the rise of autonomous device networks where different vertical requirements are summarised with an important note that each of them are under continuous evolution as well. Also, Northstream published a white paper where the different technologies for different needs in Massive IoT were discussed. Now and in future, the important requirements also include scale, density, and different performance points per device. From the business model point of view, the white papers emphasise the importance of low operational cost pointing out the problem of node level recurrent fee for IoT connectivity.

Can the two different approaches (LPWAN/PAN) now fulfil the new requirements? Obviously, using low bitrate for range excludes the viability of LPWAN technologies. Mesh has a better chance, but with its current status it is lacking scale and reliability in larger than few tens of nodes installations and when throughput requirements are raised.
So, to the main topic: what is the Wide Area Mesh then?

It is designed for the Massive IoT requirements and It includes three main characteristics that the other systems are lacking:

- It has no limit in scale and density. That is achieved by de-centralized network management

-Its main scope is in the OSI layers L2-L4 enabling use of any standard or proprietary radio on the south bound interface and IPv6 north bound maximizing the fit for purpose for a given use case and giving a standard development environment

-It has a flexibility to change at run-time the QoS per node for latency, throughput, and power.

While keeping the north and south bound APIs standard and open – the innovation can be truly unleashed in the core of Wide Area Mesh to introduce timely business driven improvements.

Too good to be true?

Let’s take an example. IDC (International Data Corporation) did a case study by interviewing Norway’s biggest utility Hafslund. In the study, the Wide Area Mesh or WAM was proven to:

-Autonomously create a network of 700 000 meters without any other infrastructure than the meters themselves – no routers, repeaters, concentrators, or base stations.

-Have 100% coverage including deep indoors and basements – with automatic and local optimization of dense metering rooms - no involvement required from any central systems or installation engineers.

-Minimum of 99.9% SLA (Service Level Agreement) on first pass installations without a clean-up process.

-Free of any recurrent fee for the connectivity for the whole life time of the meter. For comparison, if NB-IoT subscription at 25c per month per meter would be used – the saving would be about 30 million Euros for the lifetime of the metering system (15 years) and with lower KPIs for the coverage and SLA vs WAM.

-Meter to meter range in this application is up to 7km in open space conditions, which translates to about half a kilometre across urban area. That proves the power of modularity to use any radio, which in this case is a standard off the shelf Silicon Labs Sub-GHz transceiver.

For the sake of completeness, other world records with WAM so far have been

-More than 1000 radios in a cube meter. Bi-directional connectivity without a single package collision. This will be multiplied in coming releases driven by asset management density requirements. Used radio in this example is Nordic nRF52830.

-IPv6 router (what a WAM node is) continuously connected at 25uA. This was also done with Nordic nRF52830 and will be halved in future releases further improving TCO of battery operated use cases.

-Simultaneous use of different QoS devices in one network. WAM can be configured run time per node to have e.g. 100kbit/s application level network throughput at scale in one set of nodes (e.g. lighting infrastructure) and less than 10uA stand by in other set of nodes in the same network (e.g. sensors, switches, or actuators).

And what about the saying one gets what one orders? In my opinion, that is the most important part of this blog. The very successful case of Hafslund is a great example. They did not order a technology or specific standard in their tender apart from obvious requirements of IPv6 and OTA (Over The Air update). Instead, they required service levels with liabilities. Specifying a particular technology for a large-scale tender may be a billion Euros level mistake as if the technology doesn’t deliver the responsibility remains with those who specified it.

What’s your service expectation from a large scale wireless connectivity? Maybe it includes mobility and positioning or network diagnostics in addition? Contact our sales and discuss if WAM fulfils it today or if not, can it be built for tomorrow?

Teppo Hemiä, CEO @ Wirepas