IP Traceback Based Probabilistic Packet Marking and Randomized Network Coding

A novel approach for IP traceback is proposed, which combines probabilistic packet marking and network coding techniques. Dissimilar to prior PPM schemes, our approach stores random linear combinations of router labels instead of overwriting to further decrease the convergence time for attack path reconstruction. Based on the coupon collector problem with unequal probability, we also present accurate models for convergence time of the proposed approach as well as that of prior schemes. Simulation results and analysis indicate that our approach can effectively improve the traceback performance.