The main objectives of this dissertation are to develop a flood search algorithm with minimum first-derivative length (MFDL) path in circuit-switched networks and to study as algorithm of route optimization using the first-derivative length (FDL) path in hybrid-switching networks.
Circuit-switched networks for military tactical application exhibit properties inherently different from those of strategic or commercial systems. By its nature, the tactical network is highly mobile and is constantly in a state of change. The networks are frequently subject to stress situations such as dynamic traffic, damage, and jamming. Flood search algorithm is known to be an effective routing mechanism for such a tactical application, since it provides high degree of survivability and robustness. But, it is known that it has significant drawbacks with respect to the network efficiency.
In the first part of this dissertation, we consider a tactical circuit-switched grid network with a maximum of four links and two priority classes of voice traffic. The performance criterion for circuit-switched traffic is the end-to-end blocking probability. Using the minimum first-derivative length (MFDL) path, we improve the blocking probability performance of a circuit-switched network without increasing the call setup time and processor loading of the algorithm. Thus, in military applications of the technique, our scheme can improve the network efficiency over the conventional flood search algorithm.
Load balancing and routing are two of the most important issues in computer network system. The objective of load balancing is to balance the workload of the node computers by moving jobs from heavily loaded nodes to lightly loaded nodes. While the purpose of routing is to choose paths for sending messages between nodes to minimize the cost. These two research areas have been studied extensively as separate problems. However, they are closely related issues.
In the second part of this dissertation, we study the routing and optimization problem as a single problem in hybrid switching networks that support heterogeneous traffics (e.g., voice and data) and priority classes. Also, we investigate the interactions among four different classes of traffics. We assume that the hybrid interactions among four different classes of traffics. We assume that the hybrid (or integrated) network uses a time division multiplex (TDM) system with fixed duration frames. The performance criteria for circuit-switched traffic and packet-witched traffic are the end-to-end blocking probability and the end-to-end time delay respectively. This algorithm has some desirable properties compared with the gradient projection method or the flow deviation method. That is, it requires no initial specification of traffic flow, and the convexity of the objective function is not required. The numerical results show that the solution is close to the optimal solution and this method yields a substantial improvement in convergence time over the gradient projection method.
