Mr. Walter, first of all, a question of understanding: you differentiate between 'IoT wireless networks' and 'mobile networks'. Why?

Walter: Mobile networks are primarily intended for location-independent voice and data communication. They offer ever larger transmission bandwidths and very special services, such as uninterrupted handover when switching from one radio cell to another. Logically, they can also be used for IoT tasks where high bandwidth or handovers are not required. IoT wireless networks only support very low bandwidths and focus on stationary systems, such as a building sensor or a consumption meter. This technology is ideal for battery-operated low-power systems. It is therefore also referred to as low-power wide-area wireless networks (LPWA).

Let's talk about your LoRa pilot project. What tasks was LoRa intended for in your project?

Walter: The customer, an energy supplier from the heating sector in a city with several million inhabitants, wanted to integrate the current consumption data of its heating customers and certain measured values on the status of the supply network into the company's IT using a cloud. In other words, very small amounts of data in total. As a rule, one measured value per day would be sufficient for the consumption data and one measured value for each major change in the grid status. The locations of the meters to be read are predominantly basements or other locations with suboptimal radio technology.

Why isn't the 2G/3G or 4G mobile network of the established providers used for such a classic M2M task?

Walter: That was also our first proposed solution. However, due to the monthly operating costs for the SIM cards, which can add up to very substantial fixed costs with a large number of systems, the solution was practically rejected by Controlling. In the heat supply sector, there is also the fact that we are dealing with a seasonal business here. There is hardly any demand for heat in summer, but the fixed SIM card costs are still there. What's more, traditional mobile networks don't reach through thick basement walls very well.

How did you come up with the idea of LoRa as an alternative?

Walter: From a technical point of view, we have been looking at the LPWA issue for some time and have also been observing activities in neighboring countries. There were already comparable projects that obviously worked. We then looked at the LPWA provider market in Germany in the summer of 2016 and came across the IoT start-up Digimondo in Hamburg. Technically, everything fit. Digimondo also made a clear statement in 2016 that LoRa network coverage exists at our customer's location or is currently being set up. We accepted all of this as fact for the time being. In Germany, it is quite difficult to obtain reliable and meaningful financial and management data on a start-up in order to evaluate such statements. It was clear to us that we were taking a big risk, even though Digimondo presented itself in 2016 as an E.ON subsidiary that obviously had the resources to offer LoRa wireless coverage for IoT and M2M applications, at least in larger cities.

How did you proceed?

Walter: We proposed the Digimondo LoRa IoT network, which was apparently just being created, to the customer as an alternative to the existing mobile networks and offered a field test. The customer took us up on this offer. So we built a LoRa device with the necessary special interfaces. This step was possible within a few weeks with the help of our Thinglyfied technology stack and a pre-certified Microchip radio module. We then tested the first prototype together with a partner in Munich on the existing Digimondo network and achieved the desired result. The fact that this network obviously only consisted of a single or at least very few Digimondo LoRa gateways was not noticed during this first test. We then set an installation date with the customer and partner for a real field test in Q1/2017, and our customer then selected a district in their coverage area to test with several LoRa devices.

After consulting with Digimondo, it turned out that unfortunately there was no LoRa radio coverage in this district and there were no plans to change this at the time. To solve the problem, Digimondo simply sent us a LoRa gateway in an IP67 housing with an internal GPRS/UMTS radio modem and pre-installed SIM card. We had to take care of the LoRa gateway location and installation ourselves together with the customer and partner. Support was only offered in the form of paid telephone support. So we had obviously opted for a do-it-yourself LoRa infrastructure. We then selected the gateway location with our partner and the customer on site and carried out the installation ourselves. There was neither a qualified planning aid for calculating the radio cell nor any helpful gateway installation instructions. After all, there are many legal and electrical safety aspects to consider. For example, the antenna tip of a LoRa gateway on the roof of a building or on a chimney must not protrude over any existing lightning conductor.

Does that mean you have climbed onto building roofs and chimneys?

Walter: We don't, because that's not possible in terms of insurance and labor law. But the customer and our partner did climb. It was only possible to install and commission a LoRa gateway outdoors and the LoRa devices we developed in several building basements thanks to this above-average effort. This effort cost a lot of money, time and nerves, especially as we also had to build test tools to be able to measure the LoRa signal of the gateway in a basement.

What happened next?

Walter: The systems have been running for several weeks and at least provide consumption data via the Digimondo cloud. Unfortunately, too many data packets are being lost due to radio problems with the measured values for the status of the supply network, which means that we are currently unable to monitor this. In this case, we were nowhere near achieving the long ranges between device and gateway that are often emphasized for LoRa radio connections. The reasons for this could be the antenna of the Digimondo gateway, but also the gateway location itself and, above all, the single gateway. Theoretically, even the 2G/3G mobile connection to the cloud could be part of the problem.

My conclusion: LoRa radio cells must be planned and calculated in advance. The calculation should include the details of the locations and, if possible, every single wall thickness. Otherwise you shouldn't be surprised that LoRa can't even bridge one kilometer between a single gateway and the devices in a large city.

So is LoRa ultimately too expensive after all?

Walter: In March 2017, mobile communications expert Harald Naumann made an interesting calculation in a presentation: if a new wireless network provider entering the market were to equip the urban area of Hanover, for example, with LoRa gateways in order to offer any IoT customers more or less seamless wireless coverage down to the basement and parking level basement floors, gateway location rental costs of between 1.2 and 7 million euros per year can already be expected as fixed costs. Naumann calculated that each 'LoRa Base Transceiver Station (LoRa-BTS)' (equivalent to the LoRa gateway) in a typical urban environment would 'illuminate' around 0.57 square kilometers. This means that around 358 LoRa BTSs would be required for the 204 square kilometers of the city of Hanover. With monthly BTS site rental prices of between 300 and 3,000 euros (average rental price = 1,650 euros), this comes to between 1.2 and 7 million euros per year. The calculations would look similar in other cities. With a reasonably serious investment calculation, a start-up like Digimondo should have realized before entering the market that a business model as a LoRa wireless network provider is unlikely to work economically. Only for an already established provider, such as Telekom, could such an investment pay off. In the simplest case, the LoRa gateway would be an 'additional box with antenna' at an already rented location. However, as Telekom has opted for the LoRa competitor technology NB-IoT in the frequency range subject to licensing, LoRa in Germany is likely to be primarily suitable for large-scale IoT wireless networks on company premises. Here, users can install the gateways themselves. The fixed costs are then reduced to the power supply.