First steps toward scientific cyber-security experimentation in wide-area cyber-physical systems
Proceedings of the Eighth Annual Cyber Security and Information Intelligence Research Workshop
| Hi-index | 0.00 |
Critical wide-area infrastructures (CWI) need real-time monitoring and control for their reliable operation and for fast identification and resolution of anomalies within these infrastructures. Real-time monitoring and control is made possible by the use of multicast routing for fast, efficient, and reliable dissemination of status information in the CWI. The fast and efficient dissemination of status information in these infrastructures imposes delay and cost requirements on the multicast routing heuristic. To ensure reliability in the face of network failures the multicast graphs constructed by the heuristic need to have at least two paths from the source to each destination in the network. Satisfying the delay requirements of the application ensures that the data delivered to the destination is fresh. Reducing the total cost of the multicast graph helps in lowering resource usage. Additionally, having two node-disjoint paths from the source to each destination makes a multicast graph resilient to node or edge failures and improves its survivability against such failures. This dissertation presents new heuristics for constructing survivable and non-survivable low-cost delay-constrained multicast graphs and presents more extensive evaluations than have typically been used to demonstrate properties of multicast routing heuristics. The edge-priority based dynamic weight heuristic (DWH) proposed in this work constructs non-survivable, low-cost, delay-constrained multicast trees. DWH assigns dynamic weights to edges based on how close the edge is to violating the delay bound of the application. The effective-edge-costs (edge-costs influenced by edge-weights) are then used as indicators of the relative merit of including the edge in the paths from the source to the destination node. Dynamic weight disjoint path pairs (DW-DPP) heuristic is the survivable variant of DWH and constructs delay influenced (like in DWH edges are assigned varying weights based on the delay bound requirements) shared disjoint path pairs from the source to each destination node. Edges can be shared by paths to multiple destination nodes thus reducing the total cost of the multicast graph. The evaluations show that DWH and DW-DPP construct multicast graphs with 10--15% lower costs than the node-priority based heuristics that are also explored in this work.