Typical sensor nodes in Internet of Things deployments are low power, wireless, distributed over a large area, and most certainly resource constrained (usually driven off of a very low cost controller). They may contain a variety of sensors from different manufacturers, possibly using different protocols, and they are prone to failure.
But sensors need to be self-organizing, self-healing, robust, cross-platform, standards-based, and secure.
That’s the opinion of Ranjit Deshpande, CEO of K2 Inc., a startup software provider of end-to-end IoT applications, and one of the speakers at this week’s inaugural Fog Computing Conference in San Jose.
To address resource constraints you need to do software optimization. To address power consumption you need to do power management. To address environmental issues like RF interference you need to use frequency hopping. To make sensors more robust, you need to provide multiple paths for each sensor (mesh networking can allow for that). And to ensure reliability of data you need to use advanced data collection and validation – the most important part.
Software-defined radios are becoming popular in sensor networks today, he said, because there are lots of legacy sensors installed, and people don’t necessarily want to rip and replace them. An SDR can talk to legacy sensors and at the same time carry them forward with new technology, noted Deshpande.
Data coming off of sensors can be hard to understand, so that data needs to be structured, he added. And because it may be a large volume of data, at least in aggregate, you need to speed up processing of that data and address the cost of the WAN and storage. Regarding unstructured sensor data, the IoT controller can create order from the chaos – maintaining a data model for sensors, validating the accuracy of that data, and organizing and structuring that data, he said. And using rules-based aggregation and compression can reduce bandwidth requirements, he said, adding that you can enable that by applying these policies at the controller.
In terms of security, IoT networks with the proper management and control will have sensor nodes that can exclude rogue sensors from joining the network, throttle sensor data volume, and handle firmware upgrades. The controller functions as a management gateway for sensor nodes, enforces local security policy, can be managed via existing standards, configures sensor network topology (star vs mesh), and provide limited logs and alarms.
Edited by Stefania Viscusi