| With the advent of 3G technologies, mobile networks have expanded greatly. These networks have enabled the integration of new services and an enough bandwidth, allowing operators to meet the growing demand of users. This rapid evolution has led the operators to adapt their planning approach that come with new challenges. Those challenges become more important when these networks are designed to support different radio access technologies within a heterogeneous mobile network, like 4G networks. In this case, planning those networks involves other challenges, such as the considerable increase in services requests, compatibility with existing networks, management of intercellular mobility of users and a good quality of offered services. Thus, in order to create a network that allows to add or to remove components, good planning approach is needed. It is in this context, this paper aims to address the problem that occurs when the planning of a 4G network is based on an existing 3G network.;The proposed model includes: the link's costs of 4G-network equipment, the link's costs between 3G and 4G equipment, the horizontal handoff costs (intra 4G network) and the vertical handover costs (inter-3G-4G). The problem is NP-hard, a tabu search algorithm will be used. To adapt this heuristic, movements have been defined to reallocate and move nodes eNode B and SGSN in order to improve the cost of the current solution. The results of the implementation show a gap which is less then 30% between the TS results and left bound value. For others networks size, the gap is sometimes less then 1% compare to the left bound value.;The planning issue involves several sub-issue with a different level of complexity for each of them. This work mainly addresses the cell assignment problem regarding the 4G networks. Thus, the proposed solution is a mathematical model. This model has mainly two objectives: the assignment between 4G nodes, and the assignment between 3G and 4G nodes. Since the MME and SGW can be aggregated into a single gateway, an entity named SGM has been set. Thus, the model becomes a cost function involving assignments eNode B and SGSN to SGM. This model is subject to capacity constraints of SGM, and unique constraints on assignments eNode B and SGSN to SGM. |
