EE Seminar: Dense 5G Backhaul Network Planning, Including: Routing, Frequency/Power Assignment, and Fault Tolerance
Electrical Engineering Systems Seminar
Speaker: Elad Fisher
M.Sc. student under the supervision of Prof. Guy Even
Wednesday, 6th November 2024, at 15:30
Room 011, Kitot Building, Faculty of Engineering
Dense 5G Backhaul Network Planning, Including: Routing, Frequency/Power Assignment, and Fault Tolerance
Abstract
My thesis deals with the problem of designing a wireless backhaul network for dense 5G networks. The task of designing a backhaul network involves routing and assignment of frequencies as well as assigning transmission powers to links. In addition, we consider the problem of designing a backhaul network that is resilient to single-link and single-base station failures. The objectives of the backhaul design problem are to minimize the number of frequency bands used in the frequency assignment as well as minimize the number of antenna pairs (each antenna pair can support two anti-parallel links).
As the baseline algorithm for backhaul design, we consider: (1) an algorithm that performs routing based rounding of a solution to a min-cost multi-commodity flow problem, and (2) an algorithm that assigns frequencies to links using a greedy first-fit algorithm.
In the thesis, we experiment with various options to modify the linear program that solves the multi-commodity flow problem by adding integer constraints, resulting in a mixed-integer linear programming problem.
The fractional solutions of the optimization problems are rounded to obtain a routing.
Based on the mixed-integer linear program, we present a routing algorithm that reduces the number of antenna pairs used. Additionally, we introduce two iterative algorithms that alternate between routing and frequency assignment. One of these algorithms reduces the number of frequency bands used by iterating between routing using the min-cost multi-commodity flow linear program and assigning frequencies using the greedy first-fit algorithm, while adjusting link costs between iterations. The second algorithm reduces the number of frequency bands and antenna pairs used by iterating between routing and assigning frequencies using the greedy first-fit algorithm, while adding integer constraints for sets of links with high interference among them.
We present two algorithms that provide tolerance to single-link failures. We experiment with the effect of using our routing/iterative algorithms on the objectives. Our experiments demonstrate that our algorithms compared to the baseline, on average, reduce the number of frequency bands by 15% or reduce the number of antenna pairs by 5%.
In our experiments, our algorithms design a backhaul network with 91-231 base stations and 9-25 gateways within an hour.
השתתפות בסמינר תיתן קרדיט שמיעה = עפ"י רישום שם מלא + מספר ת.ז. בדף הנוכחות שיועבר באולם במהלך הסמינר
