How to create a successful ecosystem for your IoT project? Episode 6 - Choosing your network
How to create a successful ecosystem for your IoT project? Episode 6 - Choosing your network
IoT connectivity is at the crossroads of all the points in the IoT ecosystem. IoT devices cannot operate as intended without a reliable network connection, but more importantly, efficient and interoperable connectivity solutions are key to harness the rapidly growing data flux and ensure full control over smart data streams which, in turn, can give your project the edge it needs to surpass other market solutions.
Sensors, gateways, routers, applications, platforms and other systems have different requirements in terms of connectivity options based on their power consumption, range and bandwidth consumption. Options are numerous and range from Wi-Fi, Bluetooth, Cellular (3G/4G and 5G) or Low-Power Wide Area Networks (LTE-M, NB-IoT, EC-GSM, LoRaWAN, Sigfox and other). So amongst these offerings, how do you choose a good network connectivity solution for your project?
That’s the question we asked Barbara Pareglio, Executive Director for Connectivity for Aviation and Drones at GSMA, for this sixth episode of our series « How to create a successful ecosystem for your IoT project? ».
Barbara, in a few words, can you explain the mission of GSMA?
GSMA is a not-for-profit organization that represents the mobile industry as a whole. Our members are the majority of mobile operators – we have more than 750 mobile operators – but we also have people from the broader mobile ecosystem, including handset and device makers, software companies, equipment providers and internet companies, as well as organizations in adjacent industry sectors. We bring together about a 1000+ members that we represent via industry Initiatives, working groups and industry advocacy initiatives. More specifically, GSMA is working in the domain of Machine-to-Machine (M2M) and Internet of Things (IoT). We started more than 10 years ago and focused on certain areas such as utility and technology. But lately the main focus has been on the Low Power Wide Area and 5G for the Internet of things in many industries (automotive, manufacturing, aviation and drones).
We work with mobile operators and stakeholders but we also work closely with IoT start-ups and companies who want to use or implement networks for their IoT projects. For them we have created what we call the Mobile IoT innovator as a way to get close to them, understand what are their issues and have an open conversation. During the last few years we’ve had a lot of events within our Mobile World Congresses, which address start-up needs, helping them to use modules and networks in the best and most efficient way. We have also created online guides that explain how to set up certain features on the modules, also some guides to understand the technology and how it can be used.
In a nutshell, we help companies and organization understand what is at stake when they want to set up mobile connectivity and how we can —together with the ecosystem— create services that are interoperable, scalable and that can be used anywhere.
How do you approach the subject when someone comes to see you to choose the best network?
A while ago, we conducted a survey with GSMA to understand what was the main stimulus to use IoT. For 65% of the respondents —more than 1000 enterprises— the main driver was the digital transformation, but for a lot of them, it was a top down push from the management. This is never a good way. Well, IoT might be the answer but first you need to understand what is the problem you are trying to solve, where is that problem and if and how IoT can address these issues.
So, if someone comes to me saying, “I want to do something with IoT”, I would start by asking more detailed questions. What is the problem? Not the solution, the problem! Sometimes the solution can be very simple and doesn’t even involve connectivity! Say, you want to measure the temperature in a room, which solution would you choose? Do you need to install a complete measuring system and send the info to the cloud to access it from anywhere? Or do you just need a simple thermometer? It’s the same with anything. If you don’t know the problem, you can’t solve it.
Once we understand the problem, we can expand on it and think about where we want to go and how to measure it. You’ll have to set up some KPI’s for yourself in order to understand which solution can help. Starting from where most of your cost comes from to where it will actually be going. Is your problem related to energy waste, or an item lost in your supply chain? Can you make savings or reduce this cost with the help of IoT, thanks to a real time inventory for example?
Then think about what you want to achieve in the long term. And only then you can start thinking about what is the best solution for you to be able to price it.
Some people have already started to experiment with some connectivity and are overwhelmed because it’s difficult to navigate the differences of the solutions. You can test all of the solutions you want, but you can only work out which one is best for your purpose and for the long term.
Another thing, which I often underline is don’t look at the upfront cost only, look at the ROI of the choices you make. Cheap options could be more costly in the long term. You need to look at the whole solution, not just at the device itself, to get the right choices.
Some people are more interested in understanding the performances, the costs of the modules, how to use the technology or what are the differences, how we work with the standards, what are the releases, what it means to provide a standard every year, etc. and that’s important. But more importantly, you need to think about what will the future of the technology be. It’s a fast moving environment. There are revolutions, new technologies, new standards that will impact your project. It doesn’t mean that you have to throw away everything you have done to replace it, but you need to develop your solution in such a way that you ensure the longevity of the technology. To give an example, 4G has been here for a long time and now 5G is being deployed. It doesn’t mean that 4G is going to disappear overnight. 2G and 3G are still very popular (SMS, roaming data, etc.) and have been around since the 80’s or early 90’s and it’s only now that we are starting to shut down some of the 3G networks whilst ensuring standards allow for interoperability. But you need to be aware of the evolutions and how you might migrate to avoid being stuck with a technology that is going to disappear.
What are the stakes linked to the choice of a network? What do we need to think about?
There are a lot of parameters to take into account. It’s also a compromise. Everybody would love to have huge capacity, low latency, low battery consumption, cheap costs… these are all aspirations but it’s never going to happen for any of the solutions. So the point is to try to understand what you need to serve your solution, what are the limitations and what compromise you are you willing to make. Let’s look into it in detail…
First, there are performances related to the coverage. Various networks (Bluetooth, Wifi, Sigfox, LoRa, NB-IoT etc,) offer various coverage performances, frequencies and bandwidth. The differences are not huge but there are some significant ones. It’s not a question of bad or good. It really depends on what you need to do and the frequencies you need. The mobile network operator often offers huge resources on the spectrum. MNO manages those spectrum resources to serve best their customers, for example for technologies such as NB-IoT or LTE-M, which are not intense in capacity but might require more geographical or indoor coverage, then lower frequencies are more suited. On the other hand, for IoT devices that are more latency sensitive or need to transmit high volume data, then higher frequencies are more optimal. You need to consider this and be aware of the behavior of your device for different spectrum bands.
A lot of IoT devices are going to be battery powered, so the battery is going to play a very crucial role in the device performances. They all have different performances because they have been designed for different purposes. Choosing the wrong battery would compromise the performances of your device and sometimes you might need to use different batteries for different external environments. If you want to use your battery in the most efficient way, you need to understand how the battery performs and what will impact its lifetime – components, signal frequency…. There are a lot of things you can fine tune to optimize it and make it last longer.
The security is also very important – and some people tend to forget it in favor of performances.
Does the mobile operator protect his network? If your device is not behaving properly, will you be notified? A lot of developers quickly develop solutions for the IoT and if they don’t have a connection they try and retry, get stuck in a loop and end up being disconnected because they are seen as attacking the network. Whereas some networks can notify you and give you some explanations about why this is not working, (i.e the setup wasn’t made properly). That allows you to identify quickly the issues and act on them and that’s very useful during the pilot phase.
To make sure your solution is robust, you need to make sure that you catch all the exceptions and act on them. The prototyping is focusing on the right behavior, making sure the device behaves as it should, but you need to look at the wrong behaviors so you can mitigate the risks to protect the device. That’s how you recognize good designers and developers. Let’s say as per my previous example, that your battery powered device gets caught in a loop to connect to the network. Your device is unable to find a way out which drains the battery. Indeed, once you have run out of battery power and your device is dead, there’s no mitigation possible. In this instance, the developer didn’t catch all the exceptions. The network provides information about why the connection is rejected, thus the developer should have captured the error message and acted appropriately (e.g. it could be a temporary congestion, or a wrong request, etc.). It’s when you don’t act upon the response that it can all go wrong. We have had such a case in India. The buses had connected devices that reported to the warehouse. At the end of the day, all the buses started reporting at the same time as they were coming back to the bus depot. This was overloading the network and the devices where not reading the responses from the network, like at New Year when everyone is sending messages at the same time. It is not a good practice to have all the devices acting at the same time! The capacity of the network has been increased since but at the time we weren’t aware. This could have been anticipated and put in place a better scheduling for the devices. Resources are limited so regardless of which network you use, if you have a lot of devices concentrated in an area, make sure they are not sending messages at the same time and the information is distributed over time. It is important for you, for the mobile operator and for your final user to get a good service and a good experience.
And if you have network interruption, you might want to think of how the data is going to be stored and how you can retrieve it, especially if your application is dealing with sensitive information. In terms of network security, there’s no better solution than using cellular and having a SIM. The SIM provides a way to store information securely and it can be leveraged for IoT identification and authentication. Mobile network operators have put in place security measures and encryption plans. This is what we call the control plane. It identifies certain packets to be discarded, as well as preferential treatment of certain packets for which a high quality of service is required. Just be careful of the amount of data that needs to be encrypted and stored: some of the technologies offer limited space and the encryption can be low and easy to break. But generally speaking, in terms of priority services or mission critical services, cellular is the way to go. The operator manages their spectrum resources and can prioritize certain services over others based on your service level agreement (SLA). As a customer, you’ll need to request a specific service level agreement that corresponds to your security needs and will allow you to have priority over other services.
5G is enabling virtual networks, network slicing. A virtual network is created and reserved for you and only you. You can therefore manage your resources on this network and administer it according to your own priorities. That’s a good way to improve performances and better manage the criticality of the information to be received. Network slices can also be created with 4G, LTE, 5G NR (New Radio) and NB-IoT. And then you have what we call the control and the M-cord network, which allows you to create a virtualization or a segmentation of the network. It’s not following exactly the same standard but the finality is technically the same. NB-IoT for example, is by nature a separate slice: it’s a different network, with a different access. You can create separate resources for different customers of your solution so they don’t interfere with each other and you can manage your resources better and more easily. 5G is based on service enablement rather than traditional network which means that you can perform action that are confined to your own slices, such as sending messages to a group of devices or address them individually, etc.
You then have to consider other features, specific to your project. For example, do you need to upload/download? Do you need one-way communication? Two ways? How does it impact my project? Which modules are you going to use?
Some people are proud to say “I have a better consumption, coverage, I have a better this and that”… But it’s not about being better; It’s about what is a good fit for you and your project. What do I need and how do I get there? Let’s say you need to monitor a building. You can use Wifi, LoRa, Bluetooth, cellular… all of these solutions can cover your needs. Then a while later, you equip more buildings. Do you need the buildings to interact with each other? How do you do it? Are you going to install the same device everywhere? And then try to figure out how to connect them? You could maybe set up a gateway with cellular connection to connect them together? But really, rather than trying to tie the ropes later, you need to think it through before as in this instance, maybe going straight to cellular might have been a lot easier if you need to cover multiple places.
So yes, look at what you are doing but look at where you want to go and what is your target to evaluate the total cost of your solution.
What are the most common pitfalls that you encounter and that could lead to the failure of the project?
How to address the transformation
Sometimes, when the project comes from the management, the technical department is making technological decisions. Unfortunately, they do tend to go for solutions that interest them and that they like to experiment with, which might not fit in the overall purpose. There are a lot of options on the market and it takes a lot of time and effort to understand them and navigate the space. But once you invest time, resources and energy towards a specific technology, it is difficult to change the mindset. We see a lot of that. Often, the technical department even design the solution themselves. They believe it’s quicker. So they design yet another solution, which might be performing but impossible to send into production because developing an ad hoc solution costs a fortune. The perfect solution can’t be produced. They therefore have to change it all over again which is also expensive since you have to start again from scratch. So my advice here is be cautious: try not to design the solution yourself if you can, or do so only in the last resort if you can’t do otherwise. Generally speaking, try to get someone else to make it for you. There are lots of ready-made vertical solutions you can use. People, who have experience, people who failed as well, people who have seen it with someone else before. These solution providers are also able to distribute the charges across their entire customer base to reduce the costs in the long term. Creating your own network and maintaining it is not cheap and you need expertise. Especially if you develop a critical solution, you need a reliable, redundant network and 24/7 support, which you don’t get if you develop your own network. But if you do it anyway, make sure it’s interoperable.
Security is not tangible; you only feel it when it goes wrong, but then it’s already too late. And that’s another mistake people often make. Think of security first and then design your solution because retrofitting security is very expensive and doesn’t perform well.
GSMA has released the IoT security guidelines to do a self-assessment and make sure you develop trusted, reliable services that can scale as the market grows. You can also ask your supplier to do it or find some recommendations out there, like ENISA, or something else. Security of the device, the connectivity, the cloud, … there are a lot of things to check. There are also some certifications for security or accreditations on a national level, which propose tests. The more people collaborate on the issues, the quicker they can be fixed. That’s why we are collecting and sharing the issues we identify in a group where we can discuss them and find solutions together, like the GSMA Coordinated Vulnerability Disclosure Program and the GSMA Telecommunication Information Sharing and Analysis Center.
Looking at the big picture, if you had to give 3 bits of advice to someone looking at choosing the best network for them and building a successful project, what would they be?
Have a conversation with the network operator or the vendor. Several Mobile Operators provide tools and services for the IoT. They can advise you on how to use the best solution on the network to the best of their potential. Don’t develop your solution in a vacuum. They are a lot of labs and developer centers where you can test the network, the API’s, the batteries… and see how it behaves ahead of the deployment.
I work a lot with drones. These are a new type of devices that are trying to use connectivity and we experience similar pitfalls as for other IoT devices that are traditionally only using Wi-Fi. Performances are not optimized for cellular because the antenna and the frequencies are different. Performances can vary a lot depending on the type of antenna, the way you place the antenna and shield the signal, the type of module you chose and where it’s being placed. You can replace a module with another one but you need to figure out the impact it will have on the performances. It’s unlikely they’ll be the same. So we did a lot of tests with cellular, initially results showed bad performances and we realized that these bad results were due to the way the antenna was placed and also how the module were configured. Antennas are important and you need to look at their efficiency for your purpose.
It’s complicated to choose the right set of components. You have to take into account the price point according to your selling price and your ROI. And the choice is large. Thus, if you chose a module, chose one that is certified. Certified modules have been thoroughly tested to perform a certain way on the networks. GCF, the Global Certification Forum, which is responsible for certifying mobile devices that work on cellular networks in Europe has released a list of certified modules that you can choose from. You can then be sure that it’s been tested to the best standards, that the mobile network operator will accept it more readily, that its interoperable for roaming and that the module will deliver the best results.
From there you can do your own tests to make sure it works properly with the network. There is a lot of equipment and simulation tools such to help you testing the behavior of your connection for your environment. You can also buy equipment from them to test the behavior of your device on a virtual network.
From the software point of view, there’s also IoT platforms and companies such as Amazon IoT, Google IoT, Bevywise or IoTify that offer tests on the platform itself. They will allow you to make sure that the interface is built properly, that it can support a certain capacity and that the device is able to handle the requests. And then various simulators allow you to test the different elements such as the antenna, the battery, the case, etc.
Unfortunately, the simulation tool that will allow you to do all these tests on one single platform doesn’t exist yet!
Additionally, one of the complexities that you need to be aware of is that the prototype you have built and tested won’t be the same as the device you will send to production. You might create a case in 3D printing for the prototype and use a different one for the production. The plastic used for the production might be more compact than the one used for the prototype and that will impact the temperature inside the case and therefore your performances. You often have to redesign the whole thing and do more tests. The whole process can be quite lengthy, up to two years sometimes. And then you have to test everything again in the final environment where the device is going to be deployed. It takes time to find the right combination and optimization.
Connectivity is a tiny part of the project so people don’t want to spend too much time on it but it is essential, especially in industrial design. If the design is wrong, it’s not going to work.
So all of this takes time and energy but don’t be discouraged. Focus on solving the real problem to be successful. You don’t want to spend time chasing the wrong rabbit. IoT has potential but you have to do it for a reason. Focus on one problem at the time and fix it. And keep monitoring the KPIs. Costs aspects are very important, in general the closer you get to the top performances, the more expensive it becomes to step up in the improvement so, make sure to keep the costs reasonable and again, don’t lose sight of the final purpose of your solution.