Protecting IoT Networks with AI Enhanced Fog Computing

Foggy hills

Fog computing is a clever word play for computing that happens closer to IoT devices themselves than traditional cloud computing. Adding fog or edge computing to IoT device cluster makes a lot of sense in some applications as it allows to the individual sensors and devices to remain simple and low power while still providing on premises computational capacity. You would not want your self driving car to stop working just because it lost connectivity to the cloud computing node, would you? How about traffic lights?

In a previous post we discussed how mobile network operators can provide AI based IoT security services. Now it is time to look how fog computing can benefit from AI.

Undeniable Benefits of Fog Computing

In addition to up-time critical tasks like self driving cars and traffic lights an application of fog computing can be used to increase security and greatly reduce required bandwidth and pay-per-hour cloud computing time. If a large building complex has a temperature sensor in every room most of that data is going to be mundane and uninteresting. An Edge computing node can process all of those continuous data streams and only pass on averages, trends and anomalies to cloud.

There is a great security benefit to be had from connecting those thousands of inexpensive and as such probably quite insecure temperature sensors to the edge node instead of giving them direct Internet access. A more powerful edge node is easier to keep up to date and if it turns out to have a fundamental security flaw only one device must be replaced. More computational power also allows for better encryption and connection security.

How AI Can Help?

One implicit benefits of applying CAP’s Anomaly Detection AI on fog computing is detecting anomalies in sensor data. In our previous example hard coded temperature limits can be used to alert building management about problems in air conditioning or room ventilation but AI can also pick up small anomalies which may indicate someone tampering with the sensor or the sensor nearing its end-of-life.

Running edge computing node’s log files through AI can alert operators from internal and external threats. Suspicious connections from the Internet or DDoS activities are picked up quickly by the AI as well ass threats that originate from within the network. An attacker with physical access might hijack a sensor or try to act as one.

For further information from CAP

CAP Data Technologies
Tuomo Sipola, Ph.D., CEO, +358 40 753 2169