This data set is a combination of two data sets available on line. What we did in the paper is making a combination of the two data sets.  The first data set is real data set of taxi cabs’ movements in Rome [1]. For each row in this data set, we have the cab-id, date/time and GPS coordinates of the current location. The focus was on 50 cars for one minute of movements. The goal of using this data set is to simulate the movements of the mobile nodes. To simulate the delay of the MEC servers, we used real data set for servers utilisation in a data centre taken from Alibaba Cluster Trace Program [2]. The data set, after the combining the two data sets contains the following fields: V1 (car id) , V2(time of movements), V3 (position), latitude, longitude, machine name, server Id, time of taken the measurement (utilization), cpu utilization, memory utilization and the sum of cpu and memory utilization. 

Note that there are two files. We expalin the reseaon for having two files below.(this is also incldued in the paper)

Since the server data set contains a large number of rows, we make mapping between the cab’s data set and the server data set.  As aforementioned, the aim was to have a server utilisation for each movement in the mobility trace. Therefore, two types of mapping are used, namely, (1) random mapping and (2) consecutive time-based mapping. In the random mapping, for each movement in the mobility trace, a server and its utilisation were randomly selected, thus obtaining a data set that contains different servers for different time utilisation. This mapping is representative of situations such as the high-density deployment of MEC servers, high variations of MEC servers’ load or in high-speed movements. It was attempted to optimise the MEC server selection for this type of mapping. In the second mapping method, for each car movement, we select a consecutive time-based utilisation from one server. This is representative of when there are fewer MEC servers or when the mobile node is slower. 



The attached code can be used to read the data sets and generate the results.

[1] Bracciale, L.; Bonola, M.; Loreti, P.; Bianchi, G.; Amici, R.; Rabuffi, A. CRAWDAD dataset roma/taxi (v.3932014-07-17).  Downloaded from https://crawdad.org/roma/taxi/20140717, 2014.  doi:10.15783/C7QC7M.

[2] Alibaba Cluster Trace Program cluster-trace-v2018.    Downloaded from https://github.com/alibaba/395clusterdata/blob/master/cluster-trace-v2018/trace_2018.md, 2018