Multicore-based auto-scaling SEcure Neighbor Discovery for Windows operating systems

摘要

SEcure Neighbor Discovery (SEND) is proposed tocounter IPv6 Neighbor Discovery Protocol (NDP) securitythreats. However, SEND is compute-intensive. Fulfilling Hash2condition in Cryptographically Generated Addresses (CGA) isthe main heavy part of SEND. Unfortunately, CGA computationcannot see significant speed improvement when it runs onmulticore machine because CGA generation algorithm issequential. In this paper, we propose a multicore-based highperformance SEND implementation for Windows families tospeed up SEND computations. The proposed approachautomatically detects the number of processors available on amachine and creates equivalent number of working threads tocompute Hash2 condition. The parallelization mechanism isimplemented to assign CGA computation to all the cores. Whenone thread satisfies CGA Hash2 condition, the others stop. Withthe parallel approach, the speedup time has been increasedextremely by increasing the number of cores in the computingdevice. Besides the parallelization, we extend SENDimplementation to generate the key pair for CGA algorithm onthe-flyto enhance the security and to protect the privacy.