Particle Swarm Optimization Based Multi-Domain
Virtual Network Embedding
Kailing Guo, Ying Wang, Xuesong Qiu, Wenjing Li, Ailing Xiao
State Key Laboratory of Networking and Switching Technology
Beijing University of Posts and Telecommunications
Beijing, China
Email: {klguo, wangy, xsqiu, wjli, xiao_ailing}@bupt.edu.cn
Abstract—Multi-domain virtual network embedding (MVNE)
aims to embed a virtual network (VN) across multiple physical
domains while minimizing the embedding cost. A key phrase of
MVNE is VN partitioning which partitions a VN into multiple
physical domains. Since the MVNE problem is NP-hard, we
provide a heuristic VN partitioning approach named VNP-PSO
based on the Particle Swarm Optimization (PSO) to increase the
efficiency of VN partitioning. The VNP-PSO algorithm generates
a near-optimal solution of VN partitioning through the evolution
process of the particles. The simulation results show that our
proposal can increase the efficiency of VN partitioning and
decrease the embedding cost of MVNE.
Keywords—multi-domain virtual network embedding; virtual
network partitioning; Particle Swarm Optimization
I. INTRODUCTION
Network virtualization allows multiple virtual networks
(VNs) to be embedded in a shared substrate network which
may consist of one or more physical domains [1-6]. Each
physical domain belongs to an infrastructure provider
(InP).The service provider (SP) builds VNs by leasing virtual
resources from InPs. Virtual network embedding (VNE) is to
allocate physical nodes and paths to embed the virtual nodes
and links of the VNs. Although it is possible for a small size
VN to be fully embedded in a single InP, wide-area VNs
always need to be embedded in multiple InPs.
Multi-domain virtual network embedding (MVNE) allows
a VN to be embedded in multiple InPs, which creates the need
of virtual network provider (VNP). VNP is explained in [7],
which is a layer between SPs and InPs. VNP is responsible for
accumulating the limited resource information disclosed by
InPs. SP sends VNs to VNP. Then, VNP builds the VNs
according to the information accumulated. Existing solutions
for MVNE can be classified into two categories: the distributed
and centralized solutions. The distributed approaches [3, 4] can
respect the willingness of both SP and InPs. However, they
will lead to extra network transmission cost. The centralized
approach [5] based on the assumption that the SP has an
abstract view of the InPs (i.e., without knowing the information
of peering nodes), may result sub-optimal solutions. The other
centralized approach [6] investigates the feasibility of MVNE
with LID, and deals with the VN partitioning problem by using
an exact algorithm which does not further consider the
efficiency of the algorithm.
In this paper, we provide a heuristic VN partitioning
approach named VNP-PSO based on the Particle Swarm
Optimization (PSO) [8] to increase the efficiency of VN
partitioning. In VNP-PSO, there is a swarm of particles which
are randomly initialized. Each particle has its position and
velocity. The position of each particle is a possible VN
partitioning solution. The velocity leads the particle to fly to a
new position. The particles fly in the problem space by
updating their positions and velocities. A near-optimal VN
partitioning solution will be generated through the evolution
process of the particles.
II. R
ELATED WORK
The MVNE problem has been studied in [3-6] which can be
classified into two categories: the distributed [3, 4] and
centralized [5, 6].
In [3], SP needs to know at least one InP to send the VN.
An InP receives the VN and accepts a part of the VN. Then the
InP forwards rest of the VN to other InPs to complete the
whole VN. In [4], SP sends the VN to all the InPs. The InPs
volunteer to bid for the VN sub-segment they can bear. Then,
SP partitions the VN according to the quotations. The
distributed approaches [3, 4] can respect the willingness of
both SP and InPs. However, they will lead to extra network
transmission cost and do not take into account the peering
nodes information. Thus, the distributed approaches cannot
acquire optimal embedding solutions.
In [5], VN partitioning is solved using both max-flow min-
cut algorithms and linear programming techniques. The main
work is to compare the exact and heuristic VN partitioning
approaches in terms of VNE efficiency, which is based on a
highly abstract view of the underlay (i.e., without knowing the
information of peering nodes). The abstract view of the
underlay may result sub-optimal solutions. In [6], the work
takes into account the peering nodes information and
investigates the feasibility of MVNE with LID. The VN
partitioning problem is solved by using an exact algorithm.
However, the algorithm does not further consider the efficiency
of the MVNE.
Different from the above related works, we focus on the
VN partitioning problem of MVNE with LID, and provide a
heuristic algorithm named VNP-PSO to improve the efficiency
of VN partitioning.
This work was supported by the National High Technology R&D
Program of China (2013AA013502), the National Key Technology R&D
Program (2012BAH35F02), and the Fundamental Research Funds for the
Central Universities BUPT 2013RC1103.