Graph-based Approach for Efficient Tracking Area Design in Cellular Networks
| dc.contributor.advisor | Belayneh Taddesse (PhD) | |
| dc.contributor.author | Belayneh Taddesse | |
| dc.date.accessioned | 2026-04-04T11:33:52Z | |
| dc.date.available | 2026-04-04T11:33:52Z | |
| dc.date.issued | 2025-06 | |
| dc.description.abstract | Modern cellular network technologies have significantly improved in terms of number and variety of connected devices, bandwidth, latency, and spectrum efficiency, especially compared to earlier generations. Despite these developments, cellular networks face significant challenges with respect to signaling overhead, most of which is associated with location management and user tracking. This signaling cost amounts to 1/3 of the total signaling cost at the Mobility Management Entity (MME) and consumes valuable communication resources, resulting in paging discards, handover failures, increased load on the MME, and increased power consumption at the User Equipment (UE). To avoid these signaling overheads, traditionally, network operators use heuristic and distance-based approaches to partition their networks into location areas. However, these techniques often fall short in densely populated urban regions where there is high and dissimilar user mobility. On the other hand, connectivity-based clustering techniques such as graph-based clustering, if adopted properly, can offer an innovative solution to get an optimal design. This thesis aims to implement graph-based techniques to determine the appropriate number of clusters that will help to significantly reduce the overall signaling overhead associated with paging and location update messages, and hence improve the overall performance of the network. To achieve this, cell location data, paging count, handover data, and key performance indicators (KPIs) are collected from Ethio telecom’s Operation Support System (OSS). These datasets are analyzed using both distance- and graph-based clustering algorithms, namely, K-Means, Spectral Clustering, and Markov Clustering (MCL) techniques, to simulate and predict the mobility and density information of users. To evaluate the proposed approach, a real-world LTE network deployed in the city of Addis Ababa was used as a source of data and as a baseline. The evaluation results show that, MCL reduces the Paging and Tracking Area Update (TAU) signaling overheads by 24.04% and 29.85%, respectively; while reducing the total signaling overhead by 27.87%, compared to the current ground truth configuration, improving network performance and efficiency. Additionally, this thesis advances the field by adapting MCL algorithm to optimal TA design and optimization, which may find use in next-generation and other wireless networks, which can also incorporate other networks’ data to get a more accurate result. | |
| dc.identifier.uri | https://etd.aau.edu.et/handle/123456789/8047 | |
| dc.language.iso | en_US | |
| dc.publisher | Addis Ababa University | |
| dc.subject | Cellular Network | |
| dc.subject | LTE | |
| dc.subject | 5G NR | |
| dc.subject | Mobility Management | |
| dc.subject | Signaling Overhead | |
| dc.subject | Location Management | |
| dc.subject | Tracking Area | |
| dc.subject | Tracking Area Update | |
| dc.subject | Paging | |
| dc.subject | Handover | |
| dc.subject | Mobility Analysis | |
| dc.subject | MCL | |
| dc.subject | Spectral Clustering | |
| dc.subject | K-Means | |
| dc.subject | Graph-based Clustering. | |
| dc.title | Graph-based Approach for Efficient Tracking Area Design in Cellular Networks | |
| dc.type | Thesis |