90 likes | 212 Views
Network Coding-based Protection Strategy with Data Traffic Prediction Scheduling Algorithm. Shasha Liang 2011-09-15. Background. 1. Protection strategies. 2. Simulation and result analysis. 3. Conclusion. 4. Contents. Link failures. Dynamic restoration:
E N D
Network Coding-based Protection Strategy with Data Traffic Prediction Scheduling Algorithm Shasha Liang 2011-09-15
Background 1 Protection strategies 2 Simulation and result analysis 3 Conclusion 4 Contents BUPT COMTEST R&D CENTER
Link failures Dynamic restoration: when the failure occurs, dynamic recovery is used to recover the data transmitted in the links that are suffered from failures.Although there is no need extra links with dynamic restoration technique, therecovery time is usually longer, and 100% service recovery cannot be guaranteed. pre-designed protection: reserves enough bandwidth, when a failure occurs, backup paths are used to reroute the transmission and to recover the data. 1+1 and 1:1 protection is the typical technique of the pre-designed protection. The 1+1 protection need extra indicated protection links, hence the cost of which is not acceptable in most situations. Background BUPT COMTEST R&D CENTER
Protection strategies Extra link is not applied in network coding-based protection strategies. Protection against single link failure is guaranteed. We give the network model as follows, then we will introduce how to select appropriate link to transmit protection packets . BUPT COMTEST R&D CENTER
RR The path with lower traffic load are selected to transmit protection data. Max idle Define a priority for each path and high priority path will be firstly selected. PF Scheduling algorithms The strategy is that each path is used to carry protection packets in turn, that is to say, each path has equal dispatched chance. BUPT COMTEST R&D CENTER
Simulation and result analysis Encoded packets are added to a working link which will collide with the data packets. The data packet has to wait in the cache which will lead to data packet delay. Now we give some simulations as follows: BUPT COMTEST R&D CENTER
Simulation and result analysis Fig4. An unevenness data traffic Fig5. Average delay of package length BUPT COMTEST R&D CENTER
Conclusion In this paper we present a new network protection strategy against link failure. It base on network coding and data traffic prediction scheduling algorithm. When some paths have heavy load, the strategy PF can reduce more than 50% packets delay comparing strategy RR. The strategy PF also achieve the relative average, making network traffic get to a certain extent balance on each path. To fit the diverse services and networks, more precise traffic prediction model should be researched, we believe that network protection could be done more intelligently. BUPT COMTEST R&D CENTER