Case Studies

IoT Retail: Doing IoT device management deeply and properly

The Device Chronicle spoke to Peter Schenkel, CTO, FWIP, about the importance of OTA software updates and his expectations for proper IoT device management in the context of retail and ice cream vending machines in IoT retail. 

IoT device management is in Peter’s estimation is a “bizarre market”. Most solutions do not fit the needs. However, Peter goes on to say that is one of the point solutions that bridges the gap and it is the only product that meets the needs for a robust and secure OTA software updating solution. Peter came to Mender as FWIP’s previous OTA software updates provider withdrew its service from the market for IoT retail. “FWIP needed to find a new professional solution for robust and secure OTA software updates for its next generation devices in the field and Mender fitted that billing.”

FWIP’s IoT retail product

FWIP’s devices are ice cream and sorbet making machines that feature small form factors, offer many flavours and offer a very hygienic method for dispensing ice cream. “The only alternatives in IoT retail are very large machines that are hard to clean and only offer 1 to 2 flavours.” 

The FWIP large dome ice cream and sorbet machines are typically found in IoT retails locations such as bistros, restaurants, gas stations, leisure parks, and beer gardens. Remote OTA updates and remote terminal management is needed to maintain and troubleshoot these machines. “There are already 500 machines out in the field. The 500 next generation machines to be rolled out will feature Mender OTA software updates.”

The FWIP units can be found in retail units such as in Tesco supermarkets. Peter says “The FWIP is more than an ice cream dispenser, it is also a marketing unit in IoT retail, and so it needs proper IoT device management. You need to know more than just the number of passers by. You can imagine a unit whistling at the passer by, come by, we have your favourite ice cream here! It is an emerging experience with context in terms of video and audio.” 

Back office ecosystem for IoT retail

The FWIP back office ecosystem consists of a a custom asset management system with a cloud platform underneath based on AWS. There is also an ERP system that holds the master data of the device serial numbers and the customer details, then there is the operational data that is sent from the dispensers to the back end systems. The data is then consolidated with the ERP system. Then there is a portal that brings it all together based on where the machines are and what quantities and types of ice cream and sorbet are being dispensed. The more advanced data analyses are done in Tableau. Sales reports are delivered to show product dispensing and unit sales over 2 weeks periods with benchmarks on retail positioning, shop floor size and so on. 

How the OTA updates work?

With, the FWIP machines will get a single file update, with a custom built update module which puts it in the right position for the firmware. The gateway device uses Yocto Gatesgarth 3.2 for Linux. For the next generation product, the edge device is based on an integrated custom build of a Raspberry Pi computer, a sim com module antenna, engine drivers and hardware, and s new display. In the last 12 months, Peter explains, an external engineering firm custom designed the board, the firmware and the back office analytics. The new hardware is a complete PCB industrialised board which runs with 2G, 3G, 4G and 5G with a sim card. It replaces the legacy hardware set up comprising a dongle, a sim card, and a Raspberry Pi attached to the actuator and dispenser, Peter explains. 

Supplanting the mechanical approach

The traditional use case of dispensing ice cream, it is a very messy process – you have to prepare liquid mix, churn it, freeze it, expend a lot of energy and then at the end you get only one flavour from something that is big, heavy and expensive. Now with a digital machine, it’s about using IoT and data science to anticipate trends and tendencies: Which ice cream types and flavours are selling? Does the seller need to stock up? Is there something that the seller can do to prevent the ice cream machine from breaking down? Peter says it is a classical data mining “tendencies” analysis. If you see the temperature rising, will there be a tendency to sell more ice cream? If it is 3 days above 25c then you see a dramatic increase in the consumption of ice cream. But will the vendor be prepared for this? “We should keep analysing the data on a daily basis. With the data you also find that there are different flavour preferences the morning, during lunchtime and in the evenings. Ice cream is the base but other innovative offerings could come.”

FWIP has a business model for re inventing the retail experience with data analysis and IoT connectivity. An IoT connected machine with robust and secure OTA software updates, in this case an ice cream machine, can leverage data driven insights to help increase sales and margins, make the business more reactive while keeping the devices secure and fault free. Peter says “It’s something that big players such as Nestle or Unilever wish they had readily available but comes hard for them to mobilise because of the agility needed to develop fully integrated IOT solutions with hardware, software and product components. They might be able to make an ice cream but they are a long way away from being able to do what FWIP have created.”

Do device management deeply and properly

Peter has a background in consulting. He says his “DNA comes from the Fortune 500”. He also understands digital transformation and he brings that into retail. He points out that often in organisations, there is a lack of real understanding of why they should do proper IoT device management. And if they do decide to do it, they rely on capabilities within the company and rarely seek professional help and advice from outside. “It (proper IoT device management) is not a priority for them and they do not see the asset that comes from undertaking it.” Peter believes that organisations should integrate at a deeper level, just as Mender does, as you need to have full control over the operating system. Unfortunately, as Peter laments there are so many cases that are “half-baked solutions”. Typically, where a platform is set up and optimised based on whatever data is available. “But this is actually irrelevant and most achieve this, it does not address the hard topics of integrating at a very deep level,” Peter adds. 

We wish Peter and his colleagues at FWIP well in their IoT retail journey. 


Recent Articles