5 techniques for efficient IoT solutions

Feb 10, 2019 | 

The team at Movinture has had the opportunity to work on several turnaround projects – we come in with our expertise, and a balance of product, services and integration approach to successfully launch smart and connected services for our customers. We outline a few strategies, mostly technical, to highlight what are the attributes of a good IoT solution.

1. Design, architect and optimize for the normal:

When designing and developing your solution understand the frequency of your use cases and traffic patterns (e.g. data to & fro from devices) – build your solution for what happens 80 to 90% of the times. The percentage we mention should be based on your budget and timelines. For example if 90% of the packets (incoming traffic from devices) is interleaved at 100ms, and 10% is less than 10ms – design for 100ms, and have an exception path (may be even distributed between the device and the gateway & cloud) that handles packets interleaved at less than 10ms. The primary reason – if you design for the worst case, as the volume of traffic (without changing the percentage distribution between the traffic pattern) goes up – the cost also rises in proportion or faster.

2. Handle exception gracefully:

The corner cases or from the example above in 10% of the cases – providing the same processing rate or responsiveness can be sacrificed. Your solution could slow down (assuming this is not a critical requirement) but it should not fail or crash.

3. Total Cost of Ownership (TCO):

Be careful of addressing cost only as a Cost of Goods (COGS) only. You may, as an example, go for a cheap sensor or connectivity – however find out that the installs are expensive. Your monetization and profitability strategy can go all wrong if you do not take into account all costs – development + testing + install (e.g. time taken by technician to install) + support (phone call support or email support) + recurring costs (e.g. carrier charges).

4. You’re not Google Or Facebook Or ThingWorx or GE Predix:

We see this happen all the time – software engineers go crazy for the latest and the greatest (this is one of the key reasons cost spiral out of control). They will choose the latest and the greatest software and frameworks. Build for the right scale – if you ship only 100K devices a year, design for a multiple that could account for growth – for example first design and deploy for millions, and then move to “billions”. Your IoT solution does NOT most likely need the same scale as GE Predix or ThingWorx.

5. Adapt your system and process to IoT:

This is very important – you cannot build or at least you should not accept a process purely for Web Application development to work for IoT solution. Be Agile in philosophy and approach, adapt the process pragmatically for your IoT solution development.