In our standard IoT architecture, it is obvious that the bottleneck in the system is the constant flow of sensory data. This flow overwhelms your IoT dashboard vendor and results in high cellular data charges. Most of the time, the sensory data is unchanged from one transmission to the other. Your architecture is inefficient.

A smarter approach would be to use edge analytics and push off some of the processing to the edge. The following diagram shows the new and improved architecture using edge analytics:

As you can see, the vending machines are now connected to a gateway (in this case, a Raspberry Pi). Sensory data is read by the gateway instead of being sent directly into the cloud. The gateway will only send sensory data when there has been a change. This alleviates the IoT dashboard from constantly reading sensory data. Cellular data usage is also reduced.

We could build intelligence into our gateway whereby data is analyzed and messages could be sent back to the IoT dashboard. For example, if machine A seems to be constantly out of stock on Tuesdays before 4 p.m., a reminder message to refill this machine could appear on the IoT dashboard at the appropriate time.

Another intelligent function could be to compare sales from side-by-side machines with different refrigerated temperatures. Over time, one machine may dispense far more than the other, giving you an idea as to the optimal temperature for your product.

Taking the edge analytics approach, you alleviate the strain of constant network traffic and cloud processing in the system. Costs are reduced, and the system is far more efficient and up to date.

Now that we have a rudimentary understanding of edge analytics, let's take a look at the key concepts and benefits.