Network coding is a reasonable way to increase network efficiency in response to an increase of sensed data in the Internet of Things (IoT). In network coding, intermediate nodes combine packets received from neighboring nodes, transform, and transmit encoded packets that can be decoded at the destination. This scheme is based on trust among nodes. If any malicious node joins the network, it can act as an intermediate node that could fabricate encoded packets. It might be more difficult to identify the authenticity of such encoded packets since packets that are received at the destination might not originate from a single source, but be combined with several other packets originating from multiple sources. In this paper, we propose a scheme on how to detect attacked packets among the received packets at a destination and how to recover the original message from the packets including the attacked "look-like-valid" packets. This scheme shows that a destination could recover the valid message with just the received packets including some attacked packets and will result in a quite efficient performance in network coding.
Keywords: Internet of Things; attack detection; network coding; recovery.