Adaptive and Efficient Radio Access Selection and Optimisation
in a Heterogeneous Communication Environment
Queen Mary, University of London, UK
Queen Mary, University of London, UK
Abstract— In recent years, a variety of wireless network
technologies have been developed and deployed. Future mobile
communication system will have a packet switched core. The
access to the mobile communication system will not be restricted
to the mobile cellular networks but may be via other wireless or
even wired technologies. Such hybrid access can enable service
convergence, joint resource management and adaptive quality of
service. The future mobile communication system still has many
pending issues to solve. One of them is the selection of the most
appropriate radio access network when receiving a service
request. This paper addresses this issue by proposing a new
adaptive and efficient algorithm. This algorithm facilitates radio
access network selection and
application requirements, user satisfaction, gains, as well as
network resource availability and utilisation. The simulation
results show that the proposed algorithm can improve the
network performance and capacity.
Nowadays, multiple wireless network technologies are
available as commercial wireless systems. They can be
classified into three main categories: the mobile cellular
networks (2G and 3G), the wireless local area networks (e.g.
WiFi) and the wireless metropolitan area networks (e.g.
WiMAX). The overlapping of different Radio Access
Networks (RANs) creates heterogeneous communication
environments. Future moblie communication systems consider
the heterogeneous communication environments and foresee
universal access to realise service convergence, joint resource
management and adaptive quality of service . In an
environment with multiple technlogies, it is a challenge to
make the RANs cooperate with each other to achieve the above
aims. Considering that today’s wireless and fixed networks are
increaingly packet-based, one solution is to introduce a packet
switched (PS), IP-based core network, for example, the 3GPP’s
Long Term Evolution . The advantages of having a PS core
network can be: enabling intelligence at the network edge and
supporting various business models ; facilitating the
provision of diversified and flexible services which can fulfil
different user requirements ; providing a flatter network
architecture and simplifing network integration .
An internetworking architecture is presented in Fig. 1. It
includes a PS core network, which is the backbone integrating
different RANs, such as the evolved UTRAN, 2G/3G RAN,
WiFi, and WiMAX. The infrastructure of the RANs can be
maintained without modifications. The PS core can directly
connect with the evolved UTRAN, the WiFi and the WiMAX
networks. In some situations, such as for security reasons , a
gateway may be introduced between the PS core and the
WiFi/WiMAX networks. For RANs like the legacy circuit-
switched mobile networks, the connection to the PS core can
be made via an access router or a gateway, as for instance, a
Serving GPRS Service Node.
In such a heterogeneous environment, an intelligent RAN
selection system should be implemented to make the
heterogeneous communication system function in an adaptive
and efficient way. Zhu and McNair  investigate using a cost
function to solve the handover issues in a heterogeneous
environment. The cost function takes the attributes of a
network and returns a value, which represents the cost of
actual handover. The network that results in the lowest value
is selected. Koundourakis et al. present a network-based
access and interface selection system in . Their research
concentrates on the optimal usage of network resources and
provision of acceptable QoS. In , we propose a network-
centric RAN selection system, which resides in the PS core as
shown in Fig. 1. The RAN selection system collects and
updates user/terminal and network context information, and
accepts user service requests. Based on the context
information and a selection algorithm, it selects an appropriate
network for the user.
This paper describes our new access selection and
optimisation algorithm used to generate an optimal solution
for our RAN selection system. The proposed algorithm not
only selects an appropriate network, but also applies
adjustments to the network resources utilisation and
management if necessary. Section 2 describes the RAN
selection and optimisation algorithm. Section 3 presents the
performance analysis of the proposed algorithm, and section 4
concludes the paper.
RADIO ACCESS SELECTION AND OPTIMISATION
Our RAN selection and optimisation algorithm is context
Fig. 1. Internetworking Architecture
throughputs for the MUSE-VDA and the P-RASO algorithms.
When ρ is greater than 20, the network throughput introduced
by P-RASO exceeds the one of MUSE-VDA. This value
reaches about 5700 kbps when ρ is 40, while the throughput of
MUSE-VDA is about 4600 kbps.
Fig. 8 presents the values of the objective function. When
P-RASO is implemented, throughout the simulation, its value
is always greater than the value of MUSE-VDA. The greater
value of the objective function, the higher level of user
satisfaction can be obtained.
Fig. 9 compares the blocking probabilities for different
algorithms. In the simulation, the value of ?? ??_?????????
assumed as 2%. When MUSE-VDA is used, the blocking
probability starts to surge at ρ = 16 and gets over 2% at ρ =
20. When P-RASO is implemented, the blocking probability
starts to grow at ρ = 26 and exceeds 2% at ρ = 40. At the end,
the blocking probability of MUSE-VDA reaches 24% and the
blocking probability of P-RASO grow to 4%.
These simulation results show that the P-RASO algorithm
outperforms the MUSE-VDA algorithm. The use of the P-
RASO algorithm provides greater throughputs, a higher level
of user satisfaction and a lower blocking probability. By
implementing the P-RASO algorithm, the overall network
resources are effectively used to carry more traffic and the
user satisfaction is improved. The use of P-RASO algorithm
provides a good performance, because it can dynamically and
appropriately adjust the service classes of the existing users,
and allow more requests to be admitted.
In this paper, we also proposed a new adaptive and efficient
RAN selection and optimisation algorithm, P-RASO. The P-
RASO algorithm can be supplied with multiple policies and
adaptively change to use different policies. It considers to
effectively consume network resources and to maximise the
gains obtained from network and optimisation. In the future,
we will further study the performance of different policies and
policy adaptations in more complex situations.
3GPP. UTRA-UTRAN Long Term Evolution (LTE) and 3GPP System
Architecture Evolution (SAE)
Board of NGMN Limited. Next Generation Mobile Networks Beyond HSPA &
EVDO A White Paper [Online]. Available: http://www.ngmn-cooperation.com
3GPP. TR 22.978 v7.1.0 (2005-06) Technical Report, “All-IP Network (AIPN)
feasibility study (release 7)”.
P. Bosch, et al., “Flat Cellular (UMTS) Networks.” in the proceeding of the 2007
IEEE Wireless Communications & Networking Conference, WCNC 2007, pp.
Fang Zhu and Janise McNair, “Multiservice Vertical Handoff Decision
Algorithms.”, EURASIP Journal on Wireless Communications and Networking,
Vol. 2006, pp. 1-13
G. Koundourakis, et al., “Network-based Access Selection in Composite Radio
Environment.”, in the proceeding of the 2007 IEEE Wireless Communications &
Networking Conference, pp. 3877-3883
Weizhi Luo and Eliane Bodanese, “Optimizing Access in an Integrated Wireless
Network Environment”, Proceeding of 2008 Internatianl Symposium on
Performance Evaluation of Computer and Ttelecommunication Systems..
H. Holma, et al., WCDMA for UMTS – HSPA Evoluation and LTE, Fourth Edition.
John Wiley & Sons, Ltd, 2007, Chapter 8.
R. Bruno, M. Conti and E. Bregori, “Modeling TCP Throughput Over Wireless
LANs”. Proc. 17th IMACS World Congress Scientific Computation, Applied
Mathematics and Simulation, 2005.
 Shu-Jen Chen and Ching-Lai Hwang, Fuzzy Multiple Attribute Decision Making:
Methods and Applications, Springer-Verlag, 1992, pp. 465-486.
 Mainak Chatterjee, et al,, “Rate Allocation and Admission Control for
Differentiated Services in CDMA Data Networks”, IEEE Transactions on Moblie
Computing, vol. 6, NO. 2, February 2007
 Veselin Rakocevic, et al., “Performance Analysis of Bandwidth Allocation
Schemes in Multisevice IP Networks using Utility Functions”, Proceedings of the
17th International Teletraffic Congress, 2001
[Online]. Available: ftp://ftp.
Fig. 7. Network Throughputs of Different Algorithms
Fig. 8. Objective Function Values of Different Algorithms
Fig. 9. Blocking Probabilities of Different Algorithms