Network survivability is an important requirement that must be provided by network operators. In general, survivability techniques attempt to achieve two objectives, namely increasing (a) the agility of detecting and recovering from failure and (b) resource efficiency. However, these two objectives are in most cases contradictory and are difficult to achieve simultaneously. This is mainly because agility requires the use of expensive pre-provisioned protection circuits, which cannot be used by working connections.
In this project, network coding will be used to provide protection as a means of achieving survivable network operation that satisfies the above two objectives. In particular, pre-provisioned protection circuits will be shared by multiple connections where all such connections transmit their signals simultaneously and on the shared circuits. Network coding, where signals are combined before transmission, avoids the contention and collision between the signals, while simultaneously providing backup copies of the signals such that if failures take place, the signals can be easily and instantaneously extracted from the combined signals. The purpose of this project is to establish the theory and techniques for using network coding to support survivable network operation through protection. After rigorously formulating the problem of network survivability under network coding, a theory for network protection using network coding under single link failures will be introduced. The theory will then be extended to also handle the cases of multiple link and/or node failures. Stronger results for specific types of network connections will be developed. The theory will then be applied to develop implementation strategies for network coding-based protection for wireline and wireless networks, using different protocols at different layers.
The intellectual merit of this project is that it will provide the theoretical bases for the use of network coding to support protection in an agile, and resource efficient way, and will also introduce implementation methods for network coding-based protection.
The broader impact of this research is that it will provide an overarching approach for network coding-based survivability, which will extend and apply to several networking environments. The PIs expect that the developed techniques will be considered by network equipment manufacturers for adoption in their equipment and by the network operators for use in their networks, as this approach will provide cost efficiency, agility and flexibility. The project will also train graduate students in a new area of research, and in exploring novel applications of the new technique of network coding.
