When to write your IoT platform is more profitable than buying a ready-made
In late April, I told you about our sensors and mentioned a special IoT platform on which they work. It’s time to talk about this in more detail.
The platform is needed in order to ensure the management of IoT network devices of all levels and the collection of data from sensors, the storage of this information and its further processing. Yes, the market now has enough of these platforms, but they are not ready to solve the problem “out of the box”. This is either some separate pieces of the backend that are useful and would be useful to us in the work, or the same useful pieces of the frontend, but there is no such thing that everything at once and right out of the box. Even the platform closest to our needs required quite serious dopilivaniya and hiring new developers exclusively for these tasks.
We sat down, calculated the total cost of ownership and other pros and cons of using leading paid platforms, compared this with the ability to go and write our own platform. And it turned out that making our own for us is about two times cheaper, while the platform will fully comply with the stack of technologies adopted by SIBUR Digital.
Platform and components
The first thing that should have been on board such a platform was a server that would manage the wireless networks. In our case, this is LoRaWan, that is, the platform includes a LoRaWan network server, which takes full control of the base stations and devices that are in the LoRaWan network.
The second important part is the backend. It is he who carries out all operations with data that come from network sensors or from our other systems.
And the third component is the data warehouse. We use the resource developed at SIBUR Digital – the Corporate Data Node. Our Backend sends historical data from IoT devices to corporate Big Data storage.
As we wrote not so long ago, we want to be a truly data driven company, therefore it is very critical for us that this data is not just collected and stored somewhere, but that all of it is stored in exactly the same place with data from other systems . Therefore, we began to develop our platform with such a bias, the first integration with the data node has already been carried out, into which we began to add the received information.
Frontend is a convenient user interface for a single page application (web interface), which is accessible from any corporate PC.
And in order to meet corporate standards of reliability and security, the entire platform must call into the prod-circuit of the corporate network. To do this, it must be covered with all kinds of tests at the required level and correspond to the CI / CD processes adopted by us.
It turns out that the platform consists of a network server, backend, data warehouse and frontend. And now about her tasks.
What will such a platform do
The first task is to manage your wireless network. Here, management means a range of tasks from connecting base stations and their settings, working with sensors (connect, disconnect, send some additional configuration packages to them if necessary, to change the period of sending data or frequency plans) and so on.
The second task is to display to the operator all the information necessary for him. Indications from sensors and other data should be displayed for a person in a digestible form. There can be many cases of using sensors, and for each it makes sense to display information not in a standard template, but depending on what we are outputting.
The third task is administration. This is already such a service page for maintenance personnel, which will help to deal with the operation of a network of terminal devices. Sensors, in addition to the collected data about the environment or objects, transmit including their own telemetry, which can be used to track their status and performance. It helps maintenance personnel to notice in time that some kind of sensor has fallen off, or the battery is about to run out, or something else.
And so, it seems, fairly obvious things are described, nothing supernatural, right? But practice shows that all this is not in ready-made solutions. Either one or the other.
Pumped the system – pump the person
We want to give the apparatchik a full-fledged digital instrument, essentially turning it into a “digital apparatchik”. Such a worker quickly enough pumps the necessary expertise, gets a level up and becomes a digital employee. What does this mean for us in terms of talking about the platform? And the fact that in this case we should provide for this, take into account that such an apparatchik who does not have any programming skills at the start, could still create the necessary interfaces for himself, displaying the data necessary for his work.
Therefore, we made a mnemonic editor. Our sensor is added to the platform in three clicks. First: connected the sensor to the network by scanning a QR code. Manually apparatchik does not have to drive in sensor numbers, network parameters or any other data.
Next, the apparatchik takes some technological scheme in the form of a graphic file, loads it into our mnemonic editor and receives a new scheme at the output.
And, finally, it remains only to take this sensor and place it on the circuit, choosing the necessary limits of operation (for example, temperature (minimum and maximum).
And all, it turns out, in these three steps, a person created an interface for himself, with the help of which he will continue to operate the system.
Here is a clearer picture.
One of the main functions of the platform is to display notifications and statuses for staff. That is, the employee sees it all as a list of new letters in the mail client, of course, with different color priorities, depending on the criticality and the need to respond as quickly as possible (red events, yellow events). Until you react to the event, it will remind of itself with sound and visually. Again and again.
There is no such simplicity for the end user without programming skills in existing platforms. It, in principle, is understandable, because the goal of such platforms is to earn money using the time & material model of payment for improvements + fix for using the platform itself. The more improvements a client needs, the better. Because any Wishlist of such a client in terms of “And add us this, but you can still bring something out” is an additional source of income for the integrator. Alas, so far it works like this.
Therefore, we did our own. Everything went, by the way, the weather, we started in the fall of 2018, and by the spring of 2019 the main functionality was ready. I tell about this only now, and not a year ago, because in 2019 there was a long history with the registration of intellectual property rights on our platform.
Platform future
Firstly, we will actively finish it to our needs and full integration with existing systems. Many departments of our company are already interested in the product, which is good news.
Secondly, we can provide this as a Platform as a Service. We are already working on sales of our sensors and software for other companies that are also interested in this and who have come to about the same conclusion about existing paid solutions. It is clear that most of these clients are also from the oil and gas industry, here the tasks and processes are not very different, so you will have to finish something a little bit. If at all necessary.
Thirdly, not a single oil and gas company. We made a platform that can adequately work with any sensors in the LoRaWan network. That is, if you have a device that supports the open standards of the LoRa-alliance, then they can work on our platform. Therefore, there are opportunities for housing and communal services, and for agriculture, and for other industries where it is necessary to collect data from IoT sensors. Here, of course, specific improvements will be required for a specific type of sensors, yes.
I will write about the further development of the platform. If there are any additional questions, I will be happy to answer.
Vasily Ezhov
IoT product owner in SIBUR
