Resumen
In the Internet of Things (IoT), many strong constraints have to be considered when designing the connected objects, including low cost and low power, thus limited resources. The confidentiality and integrity of sensitive data must however be ensured even when they have to be processed in the cloud. Security is therefore one of the design constraints but must be achieved without the usual level of resources. In this paper, we address two very different examples showing how embedded hardware/software co-design can help in improving security in the IoT context. The first example targets so-called ?hardware attacks? and we show how some simple attacks can be made much more difficult at very low cost. This is demonstrated on a crypto-processor designed for Elliptic Curve Cryptography (ECC). A very lightweight countermeasure is implemented against Simple Power Analysis (SPA), taking advantage of the general processor usually available in the system. The second example shows how confidentiality in the cloud can be guaranteed by homomorphic encryption at a lower computational cost by taking advantage of a hardware accelerator. The proposed accelerator is very easy to implement and can easily be tuned to several encryption schemes and several trade-offs between hardware costs and computation speed.