Page 23 - ITU Journal Future and evolving technologies – Volume 2 (2021), Issue 2
P. 23
ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 2
Santa J. et al. [36] proposed a framework called MI‑ problem. However, in this solution, the authors assumed
GRATE to provide an ef icient and seamless handover of edge servers are homogeneous, that is, they have the
application services to mobile devices and support UEs’ same computational resources power, and it is not not
operations. MIGRATE leverages MEC’s capabilities and the case in practical environments; thus, this approach
the dynamic creation of virtual mobile devices to perform is limited to an environment with a homogeneous setting.
data processing and caching given the limited capabilities
of physical mobile devices, and allow mobile devices to Lee, Daewon, et al. [38] proposed an MCC proxy‑
maintain MEC services while moving to a new network based architecture to improve link performance between
mobile hosts and an algorithm to optimize bandwidth
domain with virtualisation capabilities. To provide MEC
usage. The proxy‑based algorithm includes three parts,
services closer to mobile devices, the authors considered
which are denoted as initial part, proxy election part,
edge virtualization domains, in which mobile devices
and sub‑proxy election part. In the article, the network
are deployed and whose data is updated using a local
congestion problem is solved by improving the link
access to a cluster‑based database. Then, the services are
performance using proxy as a cache server. The proxy
deployed in the cloud domain as virtual functions, and
server is selected based on four parameters, which in‑
the devices continuously pass on data to the platform
clude the type of host, the state of the host, the hardware
thanks to an SDN switch that is used as an entry point to
performance of the host and the available amount of
the wired network. The migration of services from one
concurrent connections. These four parameters are
access point to another is instantiated when the switch
detects a packet coming from the same mobile device constantly checked by the proxy manager to perform
address to a new port; when this happens, the switch proxy selection. Also, information from network layer
reports it to the SDN controller, which either re‑routes 3 is used to select the optimal access network. The
the tr ic towards the initial mobile device or requests following information is required to ind the appropriate
the preparation of a new virtualization domain to host access network: the state of the network, the hop count
a new virtual mobile device that inherits the behavior to the selected proxy, the highest capacity of the network,
and characteristics of the initial one. After that, the SDN the expected network load and the location or the depth
waits for a noti ication con irming the completion of the of network hierarchy.
action to establish a new route in the switch and send
data through the new virtual mobile device. This solution A cooperative edge caching approach to reduce de‑
can be further extended to reduce the latency of service lays in clustered mobile networks by optimizing content
migration and use an SDN multi‑controller solution. placement, small cell base stations, and bandwidth allo‑
cation in large‑scale user‑centric mobile networks based
on the stochastic network information, was proposed
4.2 Edge server and base station placement
by [29]. The proposed solution solves two problems,
solutions
the problem of content placement and that of small
cell base stations clustering. The article considered a
To solve the edge server placement problem in MCC, homogeneous mobile network with edge caching, where
Wang et al. [37] proposed a solution that uses mixed in‑ content is partially or completely stored at each small
teger programming to determine the optimal placement cell base stations after being coded into segments, the
location of the edge. The problem was irst formulated as user is served by a cluster of candidate base stations after
a multi‑objective restraint optimization that incorporates raising a content request. The mobile device seeks coded
edge servers in some appropriate locations to stabilize segments from candidate base stations in increasing
the workloads of edge servers and minimize the access order of transmission distance, if the requested content
latency. In the article, the authors considered a network is cached. Also, in case where the segments obtained
G with a set , , , … , of edge servers to be from the caches of all candidate base stations are not
2
1
3
placed in optimal places; the edge server executes suf icient to decode the segments, the closest base station
all the tasks assigned to the base stations, that is, the will fetch the rest of the bases from remote servers via
amount of requests performed by mobile users at each backhaul, and send them to the user through wireless
base station ∈ . The locations must be chosen in such transmission. If the requested content is not stored in
a way that the workloads are balanced, and the access cache, the nearest SBS will fetch the whole content from
delay reduced, taking into consideration the following remote servers.
constraints: No two edge servers share the same base sta‑
tion, each base station is co‑located with an edge server Guo et al. [39] proposed a solution to the edge placement
which will execute all mobile user task requests from the problem in order to optimally allocate workload to edge
base station. The weighting sum method was adopted clouds and minimize communication latency between the
in this solution to change the problem of edge server edge server and mobile devices. The proposed approach
placement into a signal objective optimization problem is based on K‑means and mixed‑integer quadratic pro‑
with a Pareto optimal solution which is obtained by trans‑ gramming; to solve this problem, the authors considered
forming the multi‑objective into a single optimization a mobile edge network including several base stations
© International Telecommunication Union, 2021 9