Performance costs of cryptography in securing new-generation Internet of Energy endpoint devices

Abstract

In past years, cryptography has been considered a difficult task to achieve on sensor nodes for the Internet of Energy (IoE) owing to the resource-constrained nature of 8-bit and 16-bit microcontroller units (MCUs). Previous attempts at implementing cryptographic services on wireless sensor nodes have resulted in high power consumptions, long operating times and the depletion of memory resources. Over the last decade, however, processors for the IoT and IoE have improved; with increased operating power and memory resources, longer data bus widths and low power consumption. With the improvements made to processors suitable for building IoT devices, the question remains whether endpoint nodes should still be considered capable of only supporting the most lightweight of cryptographic mechanisms. We evaluate the capabilities of a device family (Cortex-M series processors) commonly found in programmable logic controllers (PLC) to implement standard, verified software cryptographic libraries in terms of execution times, memory occupation and power consumption in order to determine their adequacy for use in smart grid applications. It was seen that the MCUs were easily capable of running standard cryptographic algorithms. However, the use of public key cryptography may still require the inclusion of a hardware crypto accelerator or the use of a secure MCU implementing public key cryptography; as the relatively long execution times seen during the operation of, for example, ECDSA could be intolerable within a real time, IoE application.