Recent years have witnessed a tremendous growth of research and development to provide mobile users with "seamless" communication through wireless media. Service provision in mobile networks is closely intertwined with network mobility management and therefore efficient management of subscriber mobility remains a challenging and an important area of research. This thesis examines management of signalling traffic in mobile networks and packet network design approaches, with specific contributions relating to i) reduction of network database signalling costs in GSM; ii) minimization of mobility signalling over the air-interface based a new technique for subscriber location management; iii) an efficient network design technique for packet communication networks with application to GPRS, based on a novel hybrid GA-Heuristic approach. A novel location management technique is proposed to reduce signalling costs in the core network, by reducing the rate of HLR-related transactions. By intelligent distribution of the HLR related signalling over a number of VLRs, the HLR is prevented from being a potential bottleneck and a single point of failure in the signalling network. Based on the analytical model developed, it is demonstrated that within the call-to-mobility range considered, the total network signalling cost as well as location management costs can be significantly reduced, compared to current strategy adopted in GSM, To address location management signalling in the access network, a number of dynamic schemes are considered and compared with the proposed adaptive multilayer technique. Through detailed simulations under various scenarios, the superiority of the proposed scheme, in terms of significant savings in the total signalling traffic (i.e. location update and paging) and ease of implementation, compared to other techniques is demonstrated . The location management scheme of GSM is used as baseline for comparisons. Finally, efficient design of packet communication networks, with application to GPRS backbone architecture, is addressed. To meet the cost and traffic requirements whilst ensuring that the delay and reliability constraints are also satisfied, a hybrid GA-heuristic approach is developed as an alternative to purely GA based and heuristic-only approaches. Under a common set of parameters, the performance of various techniques are compared and it is shown that the hybrid approach is capable of producing solutions that are superior to other typical network design methods.