Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Electrical and Computer Engineering

First Advisor

Ahmed E. Kamal


Multi-hop Wireless Networks (MWN) have drawn a lot of attention in the last decade, and will continue to be a hot and active research area in the future also. MWNs are attractive because they require much less effort to install and operate (compared to wired networks), and provide the network users with the flexibility and convenience they need. However, with these advantages comes a lot of challenges. In this work, we focus on one important challenge, namely, network survivability or the network ability to sustain failures and recover from service interruption in a timely manner. Survivability mechanisms can be divided into two main categories; Protection and restoration mechanisms. Protection is usually favored over restoration because it usually provides faster recovery. However, the problem with traditional protection schemes is that they are very demanding and consume a lot of network resources. Actually, at least 50% of the used resources in a communication session are wasted in order to provide the destination with redundant information, which can be made use of only when a network failure or information loss occurs. To overcome this problem and to make protection more feasible, we need to reduce the used network resources to provide proactive protection without compromising the recovery speed. To achieve this goal, we propose to use network coding. Basically, network coding allows intermediate network nodes to combine data packets instead of just forwarding them as is, which leads to minimizing the consumed network resources used for protection purposes. In this work we give special attention to the survivability of many-to-one wireless flows, where a set of N sources are sending data units to a common destination T. Examples of such many-to-one flows are found in Wireless Mesh Networks (WMNs) or Wireless Sensor Networks (WSNs). We present two techniques to provide proactive protection to the information flow in such communication networks. First, we present a centralized approach, for which we derive and prove the sufficient and necessary conditions that allows us to protect the many-to-one information flow against a single link failure using only one additional path. We provide a detailed study of this technique, which covers extensions for more general cases, complexity analysis that proves the NP-completeness of the problem for networks with limited min-cuts, and finally performance evaluation which shows that in the worst case our coding-based protection scheme can reduce the useful information rate by 50% (i.e., will be equivalent to traditional protection schemes). Next, we study the implementation of the previous approach when all network nodes have single transceivers. In this part of our work we first present a greedy scheduling algorithm for the sources transmissions based on digital network coding, and then we show how analog network coding can further enhance the performance of the scheduling algorithm. Our second protection scheme uses deterministic binary network coding in a distributed manner to enhance the resiliency of the Sensors-to-Base information flow against packet loss. We study the coding efficiency issue and introduce the idea of relative indexing to reduce the coding coefficients overhead. Moreover, we show through a simulation study that our approach is highly scalable and performs better as the network size and/or number of sources increases. The final part of this work deals with unicast communication sessions, where a single source node S is transmitting data to a single destination node T through multiple hops. We present a different way to handle the "survivability vs. bandwidth" tradeoff, where we show how to enhance the survivability of the S-T information flow without reducing the maximum achievable S-T information rate. The basic idea is not to protect the bottleneck links in the network, but to try to protect all other links if possible. We divide this problem into two problems: 1) pre-cut protection, which we prove it to be NP-hard, and thus, we present an ILP and a heuristic approach to solve it, and 2) post-cut protection, where we prove that all the data units that are not delivered to T directly after the min-cut can be protected against a single link failure. Using network coding in this problem allows us to maximize the number of protected data units before and after the min-cut.


Copyright Owner

Osameh Al-kofahi



Date Available


File Format


File Size

162 pages