Ongoing effort dynamic AKDs Challenge AKDs mobile vulnerable
On-going effort: dynamic AKDs Challenge: AKD’s mobile, vulnerable - area nodes may need to determine (“elect”) new AKD Requirements: • Security: AKD election in presence of malicious nodes – DKD only trusted authority – verifiability: fraudulently-elected leader eventually detected • Performance: – elected AKD that is “good” (e. g. , topologically close to area members, has high power, has high trust level) Approach: distributed leader election via expanding ring search, using data signed by DKD Dynamic Coalitions PI Meeting, July 24, 2001 1
Dynamic AKDs: distributed leader election Approach: distributed leader election via expanding ring search, using data signed by DKD • lack central location for centralized algorithm • messaage-efficient election via expanding ring search • robustness of election (as opposed to centralized optimization) wrt malicious/failed nodes • conjecture: elected leader close to optimal Dynamic Coalitions PI Meeting, July 24, 2001 2
Dynamic AKDs: Research issues • proof/verifiability of non-fraudulent election • metrics for leader election (topology, level of trust) • “quality” of elected leader: how close to optimal is elected leader? • multiple AKDs elected in partitioned or partitionable areas? • synchronous versus asynchronous algorithms Dynamic Coalitions PI Meeting, July 24, 2001 3
Conclusion • Blah blah Dynamic Coalitions PI Meeting, July 24, 2001 4
- Slides: 4