With the on-going IoT revolution, the requirements to connect billions of sensor devices has become top priority. Given that IoT end points have a broad variety of use cases, we can expect a broad variety of connectivity types to emerge in order to meet the considered IoT applications. There is few chance that this fragmented market will move towards “a one size fits all”, unlike the traditional cellular LTE market.
Figure 1: map of the main challenges for the Future Internet
A number of criteria will drive the selected method, including:
- Mobility: does the device require mobility?
- Power: does the device power has consumption constraints?
- Bandwidth: what is the level of connectivity bandwidth?
- Infrastructure: is there existing infrastructure in place that can be used?
- Coverage: does the network cover cities and beyond (including oceans and deserts)?
This paper is not intended to cover the whole map but to look at some specific issues.
Figure 2: connectivity solutions (courtesy: Nokia)
The Current situation
Over the last decade the telecommunication mobile industry as the satellite telecommunication sector have focused on delivering mobile broadband. This has been hugely successful and has driven the smartphone revolution. But today this current situation is failing to fully meet the booming needs of IoT devices where criteria such as low power, long range and low latency are likely to be much more important.
Another important issue that has not to be underestimated is that we currently face a situation where a number of IoT devices are using 2G. This is satisfying however operators are increasingly starting to look to switch off their 2G spectrum and move faster across to 4G broadband in order to struggle to meet the demand for higher capacity.
Let’s deep dive now into the main issues around IoT connectivity:
The availability of IoT spectrum is crucial. IoT will need spectrum which provides low latency and good penetration. In some use cases ensuring reliable. Quality of Service (QoS) will be crucially important. Both license exempt and unlicensed models will have a role to play.
The ability to have dedicated licensed spectrum in the sub-Ghz bands will help operators to build deployment and business cases around IoT and allow them to effectively portion the IoT segment of their business from the mobile segment. This will also allow operators to make use of the narrow bands of licensed spectrum for IoT and to avoid additional loading of their already stressed mobile networks. We are already seeing the emergence of technologies such as NB-IoT that can be deployed in traditional mobile licensed spectrum. The use of licensed bands may turn out to be the preferred option for high value IoT cases.
Figure 3: Licensed spectrum benefits
1.2 License Exempt
We are also seeing a great IoT momentum in wide area network (WAN) technologies for license exempt bands. Technologies like Lora, Sigfox, Ingenu and others are all competing to use the same blocks of sub-GHz ISM band (Industry Scientific Medical) spectrum. It is still not clear whether the ISM bands will be able to provide adequate WAN connectivity. The issues around capacity scale-up, robustness to interference and networks coverage will be highly complex to manage.
2. Managed Brand
The managed license exempt bands which are targeted only for specific IoT applications would allow a well balancing of maintaining the advantages of license exempt access against of going some ways to help manage the capacity and interference issues.
2.1 Edge Processing
Internet Protocol communications capability is key to the success of a fully interoperable IoT. IP needs to go to the end/edge device.
There are roughly two models on how IoT connectivity can be provided:
2.1.1 Devices connect directly to a base station using Low Power WAN (LPWAN). The base station typically serves a large number of devices, thus reducing costs.
2.1.2 Devices connect to a short range radio, which in turn connects to a gateway for long range communications, often via a wired network provided by someone other than the IoT device provider;
Lots of processing of data is now possible close to where the data is collected, without the need to pass data through ‘gateways’ – where there is a risk of proprietary control. 32 bit architecture is already available to do processing even at the edge. So we need an ICT infrastructure which is flexible, enabling processing to be carried out wherever it is most appropriate: at the edge or in the cloud or in between.
3. LPWAN (Low Power Wide Area Network)
The aim of LPWAN is to allow direct connectivity between devices and a base station which provides the link to the Network. A base station could serve a high number of devices, thus reducing costs.
Although we have seen a number of new players with LPWAN technologies, the market is still in its infancy. But there are initiatives aimed at driving it, like the LoRA Alliance, Sigfox and the GSMA NB-IoT Forum. LPWAN may enable low cost devices which challenge existing short range technologies.
Issues such as whether the model will use licensed or unlicensed bands are still being worked out: the early phase has been dominated by the unlicensed band technologies, which typically use the ISM 900 MHz band.
Standards bodies have been relatively slow to focus on LWPAN, but we are now seeing active standardization work at ETSI ERM (Sigfox), Weightless-N, Weightless-P, NB-IoT etc…
The success of LPWAN will depend on eg: availability of end nodes/network coverage, low cost and low power, open and accessible standards driven technology and interoperability between vendors. The early phase of LPWAN has been dominated by unlicensed use of the so called ISM band (around 900 MHz).
Figure 4: future of LPWA – four scenarios
4. Energy Efficiency of the Network
This is not only about the development of a new radio interface. An important consideration as we move towards IoT is the impact on energy consumption of the network, which is expected to increase.
New radio technologies like SDN, NFV will help us manage the new demands but at the same time we need to think about the structure of the network and its impact on energy consumption.
Rethinking the design of the network will help and we will probably need a flexible network which is capable of transferring and processing data over long distances (as at present) as well as reducing the distances and locating processing in a variety of places. This is likely to involve more computing done closer to where data is collected. This holds out the prospect of ensuring that the ICT network uses energy as efficiently as possible.