Browsing by Author "Zelalem, Tesfaye"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Design Optimization of Tracking Area in LTE Networks Using Clustering Techniques
    (Addis Ababa University, 2021-12) Zelalem, Tesfaye; Dereje, Hailemariam (PhD)
    The intense competition in the telecommunications market forces operators and service providers to require optimized and cost-effective networks. In the study of Long Term Evolution (LTE) network location management, one of the major challenges is tracing users at a low cost. In LTE networks, a Tracking Area (TA) is a logical group of cells that are used to represent the location of a User Equipment (UE) for paging. The location management signaling overhead of Tracking Area Update (TAU) versus paging is a documented performance concern in LTE networks. To respond to dynamic changes in UE distribution and UE mobility patterns, TA design must be modified over time, because the initial deployment may not be efficient in terms of signaling overhead. This thesis is focused on the planning and optimization of tracking area configurations by taking a real LTE network topology in the city of Addis Ababa, Ethiopia. One of the issues addressed in the thesis is the investigation of a baseline TA configuration for overall signaling overhead by transforming UE traces in the Call Detail Records (CDR) onto the drivable road network of the city and then rebuilding their corresponding UE traces. Two heuristic planning and optimization approaches are implemented to assign and reassign cells to new TAs to find out TAs that improve the overall signaling overhead in the network studied. The baseline configuration has 650 sites grouped into 8 TAs having an overall signaling overhead of 6.08E+08 messages. The first heuristic algorithm, which embeds the K-means clustering approach, has created 74 TAs with a minimum overall signaling overhead of 3.07E+08 messages, which is a 49.51 % improvement in total signaling overhead as compared to the baseline configuration. The total signaling overhead improvement achieved by the second heuristic algorithm using the Hierarchical Density-Based Spatial Clustering of Application with Noises (HDBSCAN) clustering approach, which created 237 TAs with a minimum overall signaling overhead of 2.46E+08, is found to be 59.54 %.