Page 1 of 18
European Journal of Applied Sciences – Vol. 10, No. 2
Publication Date: April 25, 2022
DOI:10.14738/aivp.102.11935. Deussom Djomadji, E. M., Ndje Biholong, E. N., & Tonye, E. (2022). Artificial Bee Colonies Solution for Load Sharing in a Cloud RAN.
European Journal of Applied Sciences, 10(2). 33-50.
Services for Science and Education – United Kingdom
Artificial Bee Colonies Solution for Load Sharing in a Cloud RAN
DEUSSOM DJOMADJI Eric Michel
College of Technology, University of Buea, Buea, Cameroon
NDJE BIHOLONG Evelyne Noelle
National Advanced School of Posts
Telecommunications and Information and
Communication Technologies, University of Yaoundé I, Yaoundé, Cameroon
TONYE Emmanuel
National Advanced School of Engineering of Yaoundé
University of Yaoundé I, Yaoundé, Cameroon
ABSTRACT
The integration of the cloud in the access network architecture, through the Cloud
RAN, allows to increase the network capacity thanks to the virtualization of the base
station and the sharing of cloud resources. The great innovation of this concept is
the BBU Pool present in the architecture. It is the pool that allows the term
"centralized" to be used. The processing that is carried out there is critical for the
operation of the entire network. Indeed, in a cloud based environment such as the
C-RAN, it is important that the resources are used optimally. This is where load- balancing comes in, which is defined as a technique used to evenly distribute
workloads across multiple servers or other computing resources to optimize
network performance, reliability and capacity. This paper proposes a load
balancing algorithm in the BBU pool of a C-RAN. It is based on the artificial bee
colony metaheuristic (ABC algorithm) and on the Load Balanced Min-Min Algorithm
(LBMM). All this is done to improve network performance by guaranteeing high
availability of centralized and shared resources. Its effectiveness in terms of
performance and execution time against Round Robin and MIN MIN algorithms is
studied and the results are appreciable.
Keywords: Cloud RAN, BBU pool, load-balancing, ABC (Artificial Bees Colony) algorithm,
LBMM algorithm.
INTRODUCTION
With the continuous evolution of networks, the various players in the electronic
communications sector are concerned about having networks capable of providing access to
more elaborate services such as videoconferencing, e-learning, e-Commerce, e-banking, video- on-demand and embryonic but already highly demanded services such as virtual reality,
augmented reality, v2x service, industry 4.0. The figures predict a 10 to 100 times increase in
traffic between 2020 and 2030; the multiplication of the number of devices and services; the
demand for improved user experience. The number of devices connected to the Internet is
expected to reach 50 billion by 2025. [1]
Page 2 of 18
34
European Journal of Applied Sciences (EJAS) Vol. 10, Issue 2, April-2022
Services for Science and Education – United Kingdom
To cope with this increase, mobile network operators need to become more resourceful in
increasing the capacity of their networks. Densification of network equipment has been
proposed as a solution to this problem. But this alternative is very costly as it requires the
implementation of new access points, and a relatively complex reconfiguration in the network
operation (power adjustment, cluster frequency scheme planning, traffic aggregation
equipment upgrade...).
With the integration of technological concepts such as cloud-RAN, operators are working to
solve the problem of increasing network capacity in order to contain the exponential growth of
data traffic.
The C-RAN architecture, illustrated in Figure 1, redefines radio access by separating the LTE
interface into two units: the Remote Radio Head (RRH), which is the cell, and the BaseBand Unit
(BBU), which is the "intelligent" part of the access network. The different BBUs are centralized
and virtualized in a "cloud" called the BBU pool, which is located at a certain distance from the
RRH. The two C-RAN units are then interconnected through a Fronthaul Network (FH). [2]
Figure 1 : 4G network with Cloud RAN access
However, this innovative vision of access networks presents technical implementation
challenges such as the limited capacity of the fronthaul network, the allocation of resources, the
grouping of cells, security, etc. It is in this context that our work is integrated, with the objective
of proposing an approach to improve the performance of the C-RAN network, particularly in
terms of resource sharing within the centralized BBU. This approach is the implementation of
a load balancing algorithm at the component level of the BBU Pool.
A load balancing algorithm aims to increase the utilization of low-load or idle resources,
thereby freeing up high-load resources. The algorithm tries to distribute the load among all
available resources. At the same time, it aims to minimize the makespan with the efficient use
of resources. [3]
Page 3 of 18
35
Deussom Djomadji, E. M., Ndje Biholong, E. N., & Tonye, E. (2022). Artificial Bee Colonies Solution for Load Sharing in a Cloud RAN. European Journal
of Applied Sciences, 10(2). 33-50.
URL: http://dx.doi.org/10.14738/aivp.102.11935
MATERIALS AND METHOD
The core concern of telecom operators is to efficiently respond to the continuous growth of user
data traffic. To answer this concern, it is necessary to increase the capacity of the network. The
immediate solution is to increase the density of the equipment in the network, but this raises
an obvious problem: the current architecture of access networks wastes processing resources
and their densification generates high capital and operating costs.
Indeed, the standard architecture of a base station consists of a radio module statically
connected to a processing module or baseband unit (BBU), which makes it difficult to offload
traffic from the busiest stations to less busy stations. Hence the idea of virtualizing these
processing modules and centralizing them within the BBU pool in order to manage the sharing
of processing loads between stations more easily.
Figure 2 : migration to centralized RAN
MATERIALS
Metaheuristics are iterative algorithms, having a random component and traversing the search
space via different solution generation techniques. They will iteratively move the solutions
according to different strategies in order to find an approximate value of the global solution.
Generally a metaheuristic uses information provided by its own neighborhood (it will move in
the direction that improves the result of the function), by a neighborhood of nearby solutions
or by other good solutions already found.
It is these abilities that motivated our decision to turn to the artificial bee colony (ABC)
metaheuristic in order to adapt it to the load-balancing problem; giving birth to Load Balancing
based on Artificial Bees Colony (LB_ABC).
As this work is not the first to turn to load sharing solutions, it is appropriate to present some
previous works among which are those that inspired our approach
LITERATURE REVIEW
- Tareghian and Bornaee [4] proposed a static scheduling algorithm based on the PSO (Particle
Swarm Optimization) technique. It starts with a swarm containing some particles whose
Physical board of a BBU
BBU pool with virtualized
BBUs