Meta-level network selection approach in HetNets
Küçük Resim Yok
Tarih
2026
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Springer
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
Heterogeneous networks (HetNets) introduce significant challenges for optimal network selection because of their dynamic conditions, diverse access technologies, and conflicting QoS requirements related to data rate, latency (end-to-end delay), and reliability. Existing approaches, including fuzzy logic (FL), reinforcement learning (RL), and multi-criteria decision making (MCDM), attempt to address this problem. However, their effectiveness may vary across traffic classes, network states, and evaluation scenarios. We address network selection in heterogeneous networks managed under the Software-Defined Networking (SDN) paradigm. Instead of relying on a single method, the proposed framework evaluates candidate networks using three complementary selectors: Fuzzy Logic, Q-learning, and Entropy–TOPSIS. The controller maintains a common Key Performance Indicator (KPI) snapshot consisting of data rate, latency, and call blocking ratio, and executes the selectors in parallel. If a majority of the selectors agree, the corresponding network is selected directly. In the event of full disagreement, the controller applies a controller-side meta-state-driven deterministic arbitration (CS-MSD-DA) rule based on meta-state information, including Q-table maturity, reward-trend stability, and recent selector reliability. The proposed method is evaluated using Riverbed Modeler with four access networks and four traffic classes: voice, video, sensor, and emergency. The results show that, within the considered simulation setting, the proposed controller-side meta-level framework improves decision consistency and achieves higher throughput and lower latency than both the evaluated standalone selectors and the Majority Voting with Lowest-Latency Tie-Breaking (MV-LLT) ensemble/arbitration baseline. Across different traffic classes, the proposed approach achieves throughput improvements ranging from 6 to 14%, reduces latency by up to 28.57%, and increases selection accuracy by up to 12%. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2026.
Açıklama
Anahtar Kelimeler
Entropy–TOPSIS, Fuzzy Logic, Heterogeneous Wireless Networks (HWN), Meta-Level Arbitration, Network Selection, Q-learning, QoS-Aware Decision Making, Software-Defined Networking (SDN)
Kaynak
Telecommunication Systems
WoS Q Değeri
Scopus Q Değeri
Q2
Cilt
89
Sayı
3
