Centralized Optimization of the Association in IEEE 802.11 Networks

Mohammed Amer 1, 2
2 DANTE - Dynamic Networks : Temporal and Structural Capture Approach
Inria Grenoble - Rhône-Alpes, LIP - Laboratoire de l'Informatique du Parallélisme, IXXI - Institut Rhône-Alpin des systèmes complexes
Abstract : In this thesis we study the problem of association in Wi-Fi networks. We propose solutions that allow a controller to optimize and manage in a centralized way the operations of association and reassociation. Association is expressed as combinatorial optimization problems. The proposed models consider interference between APs and are compliant with the DCF access method of the IEEE 802.11 standard. In the first model proposed we considered the case of a saturated network in which it is assumed that each AP permanently has frames to transmit. In this model, we have assumed that all the stations of the same AP have equivalent chances of transmission, ie the same number of accesses to the medium. The proposed objective function offers a good compromise between improving station throughput and equity. The numerical results obtained on realistic simulations have shown the effectiveness of this solution and show a significant improvement in WLAN performance compared to an association based on the value of the RSSI or compared to existing approaches. Subsequently, since the hypothesis of a saturated network all the time is not very realistic, we have proposed a solution that relies on real measurements such as station throughput requests and the error rates. Our solution seeks to balance the load between APs. Specifically, we seek to reduce the load of the most heavily loaded AP in the WLAN. To evaluate this load, we have proposed a mathematical model that allows to estimate the BTF "Busy Time Fraction" of an AP in any configuration (association scheme). This model is based on a Markov network. The model combined with the optimization problem allows to propose the best association. The evaluation of this solution by simulation has shown how accurate our BTF estimation, and has also shown its ability to balance the load between APs and satisfy the station throughput demands. To generalize this solution to the new versions of the IEEE 802.11 standard such as 802.11n/ac, we adapted the BTF estimation model to take into account the new improvements made by Wi-Fi on physical and MAC layers such as channel aggregation, frame aggregation and block acknowledgment. Thus, we have proposed a new metric that allows to express both the BTF of an AP and the frame aggregation rates of each of its stations. The numerical evaluation of this solution showed the advantage of the new metric compared to the BTF to improve the station throughputs and load balancing in the WLAN. It should be noted that, for the resolution of the combinatorial optimization problems formulated in this thesis, we used iterative local search heuristics. These heuristics are based on the same neighborhood structure, but the search procedures are different depending on the objective function of each model. This choice is justified by the effectiveness of local research in providing acceptable solutions in a reasonable time for complex combinatorial optimization problems
Complete list of metadatas

Cited literature [62 references]  Display  Hide  Download

https://hal.inria.fr/tel-01984703
Contributor : Abes Star <>
Submitted on : Thursday, May 9, 2019 - 11:57:10 AM
Last modification on : Thursday, November 21, 2019 - 2:16:26 AM

File

TH2018AmerMohammed.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01984703, version 2

Citation

Mohammed Amer. Centralized Optimization of the Association in IEEE 802.11 Networks. Networking and Internet Architecture [cs.NI]. Université de Lyon, 2018. English. ⟨NNT : 2018LYSE1238⟩. ⟨tel-01984703v2⟩

Share

Metrics

Record views

85

Files downloads

408