Archives mensuelles : décembre 2015

LPWAN – Tactical or Strategic move

Nowaday, every week a new network launch is announced around the world to support IoT deployments. We see many competitive network technologies fighting for this market. To name the two most prominent rolled-out, we have SigFox and LoRaWAN. But at least six other technologies are worked on and are likely to emerge as alternatives. Some are based on LTE, as NB IoT, which is the merge of the work from Nokia and Ericsson on NB-LTE and C-IoT from Huawei and Vodafone. EC-GSM is also under development to serve same purposes.

What does so many technologies in different development/deployment stages mean for the network operators and user mean?

SigFox and LoRaWan are both based on proprietary technologies, allowing a quicker roll-out, as they are already defined and being deployed in various places.

In case of SigFox, the company designed the protocol and equipment, licensing it to be used on different silicon. Sigfox remain the gatekeeper of the deployment and technology. SigFox keeps control on the network infrastructure and back end and rely on network partners to roll out the network in different countries.

LoRaWan story is a bit different, with Semtech designing the silicon for the nodes and network gateways, working with its LoRa Alliance partners to refine the network protocol. Semtech enable its ecosystem to create the building blocks allowing network deployment. Some partners will work on gateways development; some will work on the back end network to transport the data from the gateways to the back end systems using and working with the data, and network operators using these two elements to construct the network and roll it out.

Both will enable a vast ecosystem to build the end nodes connected to the network producing the data and, depending on the case, carry the commands coming from the central system to the physical world.

A great appeal of these two solutions is they rely on unlicensed frequencies allowing anybody to use them, without paying the upfront cost of frequency licenses we can see in traditional cellular networks. This characteristic allows even enterprises or group of individuals to plan collaborative network deployments, such as The Think Network or Archos. The drawback of the unlicensed frequencies usage is their potential over use, degrading the quality of service, if too many network and node are roguery deployed, as it prevent the network operators to properly design the frequency plan to support their customers.

To address this last issue, nothing prevents to have the same technology migrating to licensed spectrum. This would require some hardware adjustment, but without the need of new silicon or new software.

The main risk of these network technologies is their proprietary nature. In case a SigFox network ceases its operation, without transferring its assets to another party, it would require replacing the deployed nodes, incurring high cost to its users. In case of LoRa WAN, the risk is linked to the existence of Semtech, as it’s the sole provider of the silicon for the technology, no licensing deal being signed yet. The risk is unlikely to realize anytime soon, but the IoT is a long term technologies, with nodes deployed in the field with more than 10 years of initial lifespan.

The proprietary nature of these technologies is probably the main reasons most of the classic network operators are also looking at other technologies, such as the ones based on classical cellular solutions, as NB IoT. These will be designed by a group of companies, under 3GPP for some of them, significantly reducing the risks of their viability in the long term.

The issue with these new solutions is the delay before they would be ready for mass deployment. The initial specifications for the NB IoT are expected for 2016, with the first silicon available by the end of the year. Then the trials will start and mass industrial deployments will realistically not occur before 2018.

This gap is an issue for the incumbent network operators, as their potential customers requiring earlier deployments will likely prefer to use one available network to start their roll out, with the risk of them sticking with it, even when other networks technologies would be available, as a change would incur new development and qualification costs.

The recent network roll-out announcements by incumbent operators of unlicensed based network, as SigFox or LoRaWAN is a clear sign they want the market to be addressed now and will enable them to build competencies to be used for the future network deployments on competitive technologies.

With several competitive solutions being specified, it is hard to say which ones will get tractions and which ones will knows the limbos of dead tech. It will definitely take time to have all the solutions evaluated by the operators, then the users to identify any trend.

Some use cases of nodes connected using low power and low bandwidth with low power consumption are already known and building solutions will not wait. Therefore, it is important to carriers to provide the service today or tomorrow or being made irrelevant in this new field by the emergence of other actors.

This is why we see today incumbents deploying networks based on existing technologies and at the same time starting trials of new technologies. Orange, for example, announced a roll out of LoRaWAN network in France and at the same time indicated starting trials of EC-GSM and LTE IoT in parallel.

Some carriers haven’t yet announced any choice for their IoT related support and we can guess they would prefer not to have to support several technologies in the long run. However, this choice might be problematic as some customers could be lost to their rivals.

The LoRaWAN or SigFox can be seen as a tactical choice to fulfill today’s emerging needs and the standardized one as strategic choice for the long term support.

The reading of the fine prints for the network support and expected support in the future is important to not face an obsolete deployment in some year time and ensure the chosen technology will be supported by the operator at least for the node expected lifetime.

Multiplication des réseaux pour objets connectés

On assiste, ces temps-ci à une explosion de l’offre de réseaux à destination des objets connectés. Cette multitude d’offre est rendue possible par l’utilisation de bandes de fréquences non soumise à licence. Cela permet à tout un chacun d’ouvrir son réseau et de le proposer.

En France, Un réseau est déjà déployé, il s’agit du réseau SigFox, et trois réseaux d’accès public sont déjà annoncés. Il y a eu d’abord Bouygues Telecom, qui a annoncé le déploiement d’un réseau national pour l’année prochaine, Orange a rapidement suivi et cette semaine, Qowisio a également lancé son réseau et Archos a annoncé sa volonté de déployer un réseau collaboratif en fournissant des passerelles qui seront connecté à Internet pour transmettre les informations du terrain.

Tous ces réseaux opèrent sur la bande des 868 Mhz, qui est libre de droit, en utilisant pour les trois derniers la technologie LoRa. En parallèle à ces réseaux public, il est probable que plusieurs entreprises trouveront justifié le déploiement de leur propre réseau, en s’appuyant sur des solutions similaires.

La bande des 868 Mhz va-t-elle se retrouver saturé, un peu comme la bande des 433 Mhz, où l’on trouve beaucoup de capteurs de température, de télécommandes et autres objets grand public ?

Les objets connectés sur ces bandes de fréquences n’ont pas d’autorisations particulières à recevoir pour être utilisés en dehors de certification CE, ETSI classiques qui n’assurent que les signaux émis restent dans les limites.

Il est probable que les opérateurs télécoms, afin d’assurer un minimum de qualité de service vont regarder à utiliser des bandes de fréquences soumises à licence. Ce faisant, ils maitrisent le déploiement et le niveau de bruit sur l’ensemble du territoire qu’ils couvrent. Ce faisant, ils auront un droit de regard sur les objets se connectant sur leurs réseaux.

Cette évolution est ‘simple’ à implémenter, seule la partie RF/antenne sera à modifier par rapport aux produits déjà existant, la partie bande de base/pile logicielle restant la même.