Can Ad Hoc Routing Protocols be Shown Provably Secure?

Formal security analysis of ad hoc routing protocols has historically been lacking, normally following non-formal inspection to identify flaws and attacks. This leads to a cycle of published development, attack, development, attack, ... . Recent formal models and methods attempt to introduce rigor into the analysis approach. We discuss flaws in a previously proposed security framework based on the simulatability model, commonly used in cryptographic proofs. We show how improper application and inappropriate assumptions lead to our discovery of an attack on both the endairA and ARAN protocols, which had been previously shown to be provably secure under this model. Instead of attempting to prove security of ad hoc routing protocols, we propose a framework for analyzing attacks and vulnerabilities. This alleviates the need for a common definition of routing security or the restriction of bounding the problem domain by unattainable assumptions.