Full text
| |
| Author(s): |
João Paulo Francisco da Silva
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Computação |
| Defense date: | 2023-09-11 |
| Examining board members: |
Rafael Crivellari Saliba Schouery;
Daniel de Angelis Cordeiro
|
| Advisor: | Rafael Crivellari Saliba Schouery; Luiz Fernando Bittencourt |
| Abstract | |
Edge Computing is a computational paradigm whose objective is to bring computational resources closer to users, potentially reducing application latency. In this context, resource allocation is a challenge, since resources are distributed at the edge of the network and the demand for resources is dynamic, mainly due to the mobility of users who request these resources. In this work, we propose two sequential and truthfull auction mechanisms for resource allocation at the edge of the network considering the mobility of users. The first mechanism is a greedy algorithm for Multidimensional Multiple Knapsack Problem. The second, called 2-Phases, is a mechanism divided into two parts in which, in the first, we have a greedy algorithm for the Multidimensional Knapsack Problem, and in the second, we use a network flow algorithm. We also model a VCG-based (Vickrey-Clarke-Groves) auction mechanism for the allocation problem in Edge Computing, where the objective function is to maximize social welfare. We chose an auction mechanism from the literature to compare with the ones we modeled in this work, and then we evaluated them through simulations and a dataset of bus lines in São Paulo - Brazil, aiming to simulate the mobility of users requesting resources at the edge of the network. The results show that the mechanisms reach values close to the optimum, and also manage to allocate resources to users in a way that they have a low latency, even in mobility scenarios (AU) | |
| FAPESP's process: | 21/06282-4 - Game Theory applied to resource allocation and pricing in Edge Computing |
| Grantee: | João Paulo Francisco da Silva |
| Support Opportunities: | Scholarships in Brazil - Master |