A survey on edge and fog nodes' placement methods, techniques, parameters, and constraints

Abstract

Within Edge and Fog computing, edge and fog nodes must be optimally located at the network edge to minimise the network's overall latency. This survey addresses all aspects of these nodes' placement problems. Literature on edge and fog nodes' placement is collected from reputable databases (IEEE Xplore digital library, Scopus, ScienceDirect, and Web of Science) using a search query. Manual search using keywords and the snowball method is also used to get as many related papers as possible. According to defined inclusion criteria, retrieved documents are filtered to 64 articles for eight years (2015–2022). Depending on the optimisation method used, literature is classified into six categories. The first relies on Integer programming, accounting for 20.3% (13/64). The second category depends on heuristic and metaheuristic methods, accounting for 20.3% (13/64). The third category depends on hybrid methods between the two aforementioned categories accounting for 18.7% (12/64). Forth category depends on clustering methods, accounting for 11% (7/64). The fifth category depends on reinforcement learning, accounting for 6.3% (4/64). And the final category depends on the hybrid methods between two or more methods mentioned above, accounting for 23.4% (15/64). Papers have been analysed to get information like the optimisation problem, the method used for solving it, considered parameters, objectives, constraints, implementation tools, and evaluation methods.