D(e|i)aling with VoIP: Robust Prevention of DIAL Attacks

We carry out attacks using Internet services that aim to keep telephone devices busy, hindering legitimate callers from gaining access. We use the term DIAL (Digitally Initiated Abuse of teLephones), or, in the simple form, Dial attack, to refer to this behavior. We develop a simulation environment for modeling a Dial attack in order to quantify its full potential and measure the effect of attack parameters. Based on the simulation’s results we perform the attack in the real-world. By using a Voice over IP (VoIP) provider as the attack medium, we manage to hold an existing landline device busy for 85% of the attack duration by issuing only 3 calls per second and, thus, render the device unusable. The attack has zero financial cost, requires negligible computational resources and cannot be traced back to the attacker. Furthermore, the nature of the attack is such that anyone can launch a Dial attack towards any telephone device. Our investigation of existing countermeasures in VoIP providers shows that they follow an all-or-nothing approach, but most importantly, that their anomaly detection systems react slowly against our attacks, as we managed to issue tens of thousands of calls before getting spotted. To cope with this, we propose a flexible anomaly detection system for VoIP calls, which promotes fairness for callers. With our system in place it is hard for an adversary to keep the device busy for more than 5% of the duration of the attack.