Hash-Based Sequential Aggregate and Forward Secure Signature for Unattended Wireless Sensor Networks

Unattended Wireless Sensor Networks (UWSNs) operating in hostile environments face great security and performance challenges due to the lack of continuous real-time communication between senders (sensors) and receivers (e.g., mobile data collectors, static sinks). The lack of real-time communication forces sensors to accumulate the sensed data possibly for long time periods, along with the corresponding signatures for authentication purposes. Moreover, non-real-time characteristic of UWSNs makes sensors vulnerable especially to active adversaries, which compromise sensors and extract all data stored in them. Hence, it is critical to have forward security property such that even if the adversary can compromise the current keying materials, she cannot modify or forge authenticated data generated before the node compromise. Forward secure and aggregate signatures are cryptographic primitives developed to address these issues. Unfortunately, existing forward secure and aggregate signature schemes either impose substantial computation and storage overhead, or do not allow public verifiability, thereby impractical for resource-constrained UWSNs. In order to address these problems, we propose a new class of signature schemes, which we refer to as Hash-Based Sequential Aggregate and Forward Secure Signature (HaSAFSS). Such a scheme allows a signer to sequentially generate a compact, fixed-size, and publicly verifiable signature at a nearly optimal computational cost. We propose two HaSAFSS schemes, symmetric HaSAFSS (Sym-HaSAFSS) and Elliptic Curve Cryptography (ECC) based HaSAFSS (ECC-HaSAFSS). Both schemes integrate the efficiency of MAC-based aggregate signatures and the public verifiability of bilinear map based signatures by preserving forward security via Timed-Release Encryption (TRE). We demonstrate that our schemes are secure under appropriate computational assumptions. We also show that our schemes are significantly more efficient in terms of both computational an- d storage overheads than previous schemes, and therefore quite practical for even highly resource-constrained UWSN applications.