CCSS: An Architecture for Joint Integration of Inter and Intrasession Network Coding in Lossy Multihop Wireless Networks



Maximizing the throughput while achieving fairness among the flows is one of the fundamental research problems in multihop wireless networks. Achieving this goal is not a straightforward extension from wireline networks due to the special features of wireless links. The main distinguishing features of wireless links are the lossy behavior, broadcast nature, and interference. Network coding has emerged as a promising approach to enhance the performance of wireless networks. Two different types of network coding are used to achieve different objectives. Intrasession network coding (IANC), where only packets of the same flow or session are coded together, can be used as a smart way of providing redundancy to tackle the lossy behavior of wireless links. On the other hand, intersession network coding (IRNC) exploits the broadcast advantage of wireless links by mixing different flows at intermediate nodes to resolve bottlenecks. IANC and IRNC have been studied separately in most of the current literature. While it is not straightforward to combine both IANC and IRNC, the main objective of this proposal is to provide a framework to study and deploy IANC and IRNC jointly under different wireless network settings.

Intellectual Merit: State-of-the-art IRNC protocols fail to work under moderate channel loss rates of 20%. On the other hand, IANC fails to resolve bottlenecks, because coding is done within each flow or session separately. Under very limited settings, such as two-hop relay networks, the two approaches can be used jointly. Simulation results showed that under this simple setting, the throughput could be increased three fold while maintaining fairness compared to using IANC or IRNC separately. However, even under this simple setting, the complexity of the joint optimal scheme is super-exponential. It is crucial to deploy the joint mechanism in multihop wireless networks while maintaining moderate complexity that the nodes can handle. Specifically, the goals of the proposed research are as follows: (1) Understanding the fundamental limits for using IANC and IRNC jointly under multihop wireless network settings and for multiple multicast and broadcast sessions; (2) Creating a generic formulation for throughput maximization and fairness that jointly optimize routing and link scheduling for multihop wireless networks when IANC and IRNC are used jointly; (3) Developing low complexity and distributed algorithms for joint IANC and IRNC with theoretical performance guarantees. The distributed algorithms should satisfy the requirements of the deployment. Practical deployment requires protocols that do not need to estimate or know the channel conditions. Also, practical deployment requires distributed algorithms that can be integrated with transport protocols, such as TCP, with minimal changes; (4) Deploying the different distributed algorithms under different multihop wireless network systems to determine the most suitable algorithm for the specific setting. These systems include mesh, sensor, WiMax, and hybrid networks.


Broader Impact: The proposed research directions presented in this proposal will contribute to the broad wireless networking area. Many fundamental questions on when and how to use network coding will be revised and addressed. We believe that the answers to these questions will make network coding more practical and attractive for wireless networks. The joint integration of IANC and IRNC will strengthen the understanding of the fundamental capacity limit of lossy wireless networks and the methods to approach it. The proposed research project will also focus on integrating network coding with minimal changes to existing protocols, such as TCP. Therefore, we believe that the results of this research will contribute to the Internet standards and RFCs. We also plan to deploy our results on different wireless networks in the city of Philadelphia and Temple University campus as part of the Digital Philadelphia initiative. As technology advances, the requirements for wireless networks change. Our research objective is not to only consider current wireless networks capabilities, but to also develop solutions that future and more powerful networks can use. The PIs are planning to publish the research findings in top IEEE/ACM journals and conferences.


 



Contador    Since July 2003