RSVP-CN: ONE IMPROVED RSVP MECHANISM ADAPTED TO

COGNITIVE NETWORKS*

Wei Qi

School of Electronics and Information Engineering, Beijing Jiaotong University, Beijing, China

Xiaomin Zhu

School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing, China

Keyword: Cognitive Network, QoS, Resource Reservation, Adaptive Borrowing.

Abstract: As an effective measure to ensure an end to end QoS, RSVP mechanism has been widely used in the QoS

demanding applications. However, owing to the higher bandwidth requirements and the more complex

circumstance in the networks, the current RSVP Mechanisms cannot satisfy the user's increasing

requirements for Internet’s QoS. The Cognitive Network has been proposed as a solution for building

improved network architecture. On the basis of analyzing the existing resource reservation mechanisms and

the features of cognitive network, this paper combines RVSP mechanism and the features of cognitive

networks, and accordingly proposes one Improved RSVP Mechanism named RSVP-Cognition Network

(RSVP-CN) to optimize the end-to-end QoS. Comprehensive simulations show that the proposed RSVP-CN

mechanism efficiently improves the end-to-end QoS in the networks.

1 INTRODUCTION

With a rapid development of Internet technologies

and the increasing requirements on communication,

the new network services are emerging continually.

And Internet traffic is growing at an exponential

rate. Also, it has been driven by the increasing

number of users and the introduction of new

demanding applications. Clearly, the traditional

network architecture that has been built in the past

few years cannot offer an acceptable level of QoS to

this type of applications. To address this problem,

the concept of Cognitive Networks (Thomas R.W.,

2005 and Fortuna Carolina, 2009) was proposed as a

solution for providing end-to-end QoS guarantees

over the Internet. But At present, the research

involved with issues of mechanisms adapted to

cognition network is still in its infancy. As

the previous situation, the RSVP mechanism

* This work is partially supported by the grants of National

Natural Science Foundation of China (NSFC) with contract

number 60773033, and National High Technology

Research and Development Program Plan of China (863

Plan) with contract number 2009AA01Z211.

(Ahmad Belhoul, 2009), which is one of the most

commonly applicable resource reservation

mechanisms, has difficulties to cope with the

growing of Internet traffic Aforementioned. The

current RSVP mechanism uses the Resource

Reservation to guarantee some important network

services’ level of QoS. And based on RSVP

mechanism, Differentiated Services (DiffServ)

(Spiridon Bakiras, 2004) was proposed for achieving

a more suitable Resource Reservation for different

types of network services. However, resource of

network which has been reserved in both RSVP and

DiffServ cannot be used by other network services

or changed dynamically according to the Internet

circumstance changing all the time. The current

RSVP mechanism wastes the limited resource and

cannot satisfy the user's increasing requirements for

the end-to-end QoS. Furthermore, it cannot use the

features of Cognition Networks, such as automatic

learning, intelligence and dynamical adaptation.

To solve the above problems, this paper

introduces an improved RSVP mechanism adapted

to cognitive networks named RSVP-Cognition

Network (RSVP-CN). In this mechanism, to make

257

Qi W. and Zhu X.

RSVP-CN: ONE IMPROVED RSVP MECHANISM ADAPTED TO COGNITIVE NETWORKS.

DOI: 10.5220/0003268902570262

In Proceedings of the Twelfth International Conference on Informatics and Semiotics in Organisations (ICISO 2010), page

ISBN: 978-989-8425-26-3

fully use of the advantages in cognitive network, the

features of Cognitive Network (e.g. automatic

learning, intelligence and dynamical adaptation) are

introduced into the traditional RSVP mechanism.

The RSVP-CN mentioned in this article is adapted

to Cognition Networks and can increase the

utilization of Network Resource. Thus, the

RSVP-CN mechanism provides the higher level of

QoS.

The rest of this paper is organized as follows.

The Literature Review is given in the section2. The

related knowledge is introduced in the section 3. The

detail of RSVP-CN and the reasons why this paper

introduces RSVP-CN are given in section 4. Section

5 reports a set of simulation experiments. Section 6

draws some conclusions.

2 LITERATURE REVIEW

Many Reservation Resource mechanisms have been

proposed to solve the problems caused by the rapid

development of internet and the increasing

requirements from users. Nowadays some of them

have been mature enough.

In BGRP (border gateway reservation protocol,

(Ping P. Pan, 2000 and, Mohit Chamania, 2009)),

the network is divided into multiple domains, and

each domain can run their different resource

reservation protocols. And then, all kinds of

resources, status and information are stored in the

domain tree structure in accordance with the core

node. At finally, the router in backbone network set

aside information from each domain as a whole to

grasp the resource usage of entire network. Similar

to the former, DARIS (dynamic aggregation of

reservations for internet services, (R.Bless, 2004))

minimizes the resulting overhead, while gathering

the information of resource reservation in network.

The studies on resource reservation mechanisms

in wireless network also exist much. In the All-IP

wireless network, the cell-based switching controller

resource reservation mechanism (Bongkyo Moon,

2004) can ensure that QoS wouldn’t debase while

the mobile terminals move. Besides, Jiongkuan Hou,

Yuguang Fang (2000) proposed a mechanism named

Mobility-based channel reservation scheme for

wireless mobile networks. The mechanism uses

Impact Curves to describe the interference in the

adjacent cell caused by a new service, and design a

forecast and resource reservation algorithms based

on the interference, finally solve the problem that

service outages caused by the switch between cells.

Moreover, Task dividable based reservation for

grid computing mechanism (PU Jing, 2008) divided

resource in network into multiple sub-sets aside to

improve system throughout. A Pre-Reservation

Technology Based on RSVP (Guo Meng, 2004),

which applied to distance learning,

video-on-demand or other real-time multimedia

services, gets right to use the resources in advance

reservation. So from the reservation to the real

beginning, this mechanism allows to set aside

resources to be adjusted or canceled. Such measures

improve the flexibility for the use of network

resources, but they are still complex.

Although all the mechanisms above have a

significant effect on improving network resource

utilization and reducing rejection rate, the current

studies rarely pay attention on borrowing and

deployment the resources that have been set aside.

This paper introduces the feature of cognition

networks into RSVP, and then proposes the

RSVP-CN that provides QoS differentiated services

for the heterogeneous users and allows services with

a higher QoS level take up temporary operations in

the idle reservation resource for other services.

3 RELATED KNOWLEDGE

3.1 Cognition Networks

Cognitive networks (Thomas R.W., 2005 and

Fortuna Carolina, 2009) has been an important issue

in the field of future communications, Cognitive

behavioural model (Thomas R.W., 2005 and Fortuna

Carolina, 2009) is shown in Figure 1: in a cognitive

process that can perceive current network

conditions, and then plan, decide and act on those

conditions and can learn from these adaptations and

use them to make future decisions, all while taking

into account end-to-end goals. Cognitive technology

endows communication entities with an ability of

being cognitive for the surrounding environment and

makes them change dynamically, intellectively and

self-adaptively according to the surrounding

environment.

ICISO 2010 - International Conference on Informatics and Semiotics in Organisations

258

Figure 1: Cognitive behavioral model.

So, benefited from the features of Cognitive

Networks, RSVP-CN can perceive the occupied

reservation resources, and then decide to borrow the

idle reservation resources. After acting, the change

that services with a higher class of QoS take up

temporary operations in the idle reservation resource

for other services feedbacks to the environment. And

then, a new circuit begins again.

3.2 RSVP Mechanism

RSVP (Ahmad Belhoul, 2009) can guarantee

end-to-end QoS by exchanging PATH and RESV

messages to reserve bandwidth of the routes. In

RSVP, a source node sends PATH message to keep

the state of reservation and route. Then, when the

terminal which needs to reserve resources receives

the PATH message, it feedbacks the RESV message.

If reservation fails, the router returns RESVERR

message in order to inform terminals and source

nodes for processing. And in RSVP, reservation

successes if the path satisfies the bandwidth request.

But when idle resources don’t exist, the router

returns error message, the process is shown in

Figure 2.



Figure 2: Reservation Process in RSVP.

While implementing QoS differentiated services,

the PATH messages can carry the information that

identifies the class of services’ QoS. And this paper

adds one module to identify the classes of QoS and

decide whether to borrow the resource that has been

reserved to other services.

4 THE CONSTRUCTION OF

RSVP-CN

4.1 Basic Ideas

The current RSVP mechanism cannot satisfy the

requirement of the acceptable level of QoS from

users and isn’t adapted to Cognition Networks.

Hence, to solve the problems(e.g. resource wasting,

low level of QoS), this paper proposed RSVP-CN

which has a cognitive process that can perceive

current network conditions, and then plan, decide

and act on those conditions and can learn from these

adaptations and use them to make future decisions,

all while taking into account end-to-end goals. And

in cognition networks, RSVP-CN uses the features

(e.g. automatic learning, intelligence and dynamical

adaptation) to identify the class of services’ QoS and

discovers occupied reservation. With the accurate

judgment toward the circumstance of Internet,

services with a higher QoS class can take up

temporary operations in the idle reservation resource

for other services. And the classes of QoS are

differentiated according to the variety services’

classes. As an improved RVSP mechanism, the

advance compared with RSVP reflects in two

aspects below:

1) RSVP-CN provides QoS differentiated

services for the heterogeneous users.

2) RSVP-CN adds a module in the process of the

RSVP, which achieves the aim of identifying

the classes of the QoS and borrowing the

reservation resource to the services with higher

QoS class.

4.2 QoS Differentiated in RSVP-CN

In the network, heterogeneous services exist.

Referring to their different importance, the services

can be differentiated to several classes. In contrast to

the traditional RSVP, the DiffServ (Spiridon

Bakiras, 2004) whose benefit is scalability adopts

this method. In RSVP-CN, as the balance between

RSVP-CN: ONE IMPROVED RSVP MECHANISM ADAPTED TO COGNITIVE NETWORKS

259

efficiency and accuracy is taken into account, QoS is

divided to three classes, which are identified as high,

middle and low. And this mechanism marks them

with 1, 2 and 3. Thus, the mapping between services

and classes of their QoS is shown in Table 1.

Table 1: Mapping between services and classes of their

QoS.

QoS Class Services Class

1 Low

2 Middle

3 High

RSVP-CN allows services with a higher QoS

class take up temporary operations in the idle

reservation resource for other services.

4.3 Cognition Module

In RSVP-CN, when a new service requests for

reservation resource, RSVP mechanism judges

whether build the reservation according to the

bandwidth. Of course, while the route satisfies the

reservation’s requirement for bandwidth, it can be

allocated bandwidth for reservation. But in

Cognition Networks, the communication entities can

perceive current network conditions. Hence, the

source nodes, routers and terminals can search the

idle reservation resources, and then decide that

whether to borrow them or not. So this paper

proposed one improved mechanism named

RSVP-CN that adds a cognition module in the

process of RSVP. And the module is shown in

Figure 3.

Compared with the RSVP mechanism,

RSVP-CN adds a process named Cognition Module

which has two functions below:

3) Identifying the classes of QoS differentiated

services.

4) Judging whether to borrow the reservation

resource to the services with high QoS class.

The added module collects the information

perceived by source nodes, terminals and routers,

and then implements the process that allocating the

idle reservation resource requested by services

whose QoS class is lower than new reservation

request. Besides, new service that has the same QoS

Figure 3: Reservation Process in RSVP-CN.

class with previous one would be refused. And after

the new service finishes occupation, it returns the

resource to previous reservation. Moreover, if some

services are so important that reservation can’t be

borrowed their classes of QoS would be identified as

the highest. So, RVSP-CN would be little possible to

have a bad effect on the communication of important

services (e.g. military communication).

5 SIMULATIONS

RVSP-CN mentioned in this paper was simulated

with the well-known OPNET network simulator.

This paper chooses the simulated network with a

typical topology shown in Figure 4 below.

Figure 4: Simulated topology.

The widths of access links from the sender and to

the receivers were set to 100Mb/s. The time that

simulation takes is 150s. Additionally, a simulation

for comparison is made with RSVP [10], which is

the original mechanism optimized in this paper. And

in the topology that this paper chooses, Source node

ICISO 2010 - International Conference on Informatics and Semiotics in Organisations

260

A and Source node B send reservation requests for

services with different QoS level in a certain rate.

And this paper measured number of successful

requests, which reflects resource utilization and

network efficiency. The simulation result is shown

in Figure 5 below.

Figure 5: Number of successful requests.

In Figure 5, blue line shows the number of

successful requests when topology used RSVP-CN.

And the other line reflects the number of successful

requests while using RSVP. Through comparison we

can conclude that the improved mechanism has a

significant increase in the number of requests for

accept services’ reservation resource. Hence,

RSVP-CN can provide higher network resource

utilization, reducing rejection rate and network

efficiency.

Additionally, in order to reflect the change in

QoS, this paper measured end-to-end delay between

Router A and Router B, which is an important index

for QoS. The simulation result is shown in Figure 6

below.

Figure 6: Delay (seconds).

In Figure 6, blue line shows end-to-end delay

when topology used RSVP-CN. And the other line

reflects end-to-end delay while using RSVP.

Through comparison we can conclude that although

the blue line generally is over the red line, the

difference is not large. Hence, RSVP-CN don’t

recede QoS, but it has increased the number of

requests.

6 CONCLUSIONS

According to cognitive networks being capacity of

perceiving the state parameters of internet’s

environment self-learning and adaptively. This paper

proposed one improved RSVP mechanism adapted

to cognitive networks – RVSP-CN. The proposed

mechanism combined with features of cognition

networks is an effective way to provide reservation

resource in the internet.

Today introducing the concept of cognition

networks to current mechanisms is still in its

infancy. But in the future there will be more relevant

studies in the field. And the mechanisms adapted to

cognition networks will involve with more aspects in

the network, such as server load balance, the

existing prediction algorithms, even for the

re-classification and definition for QoS indicators.

REFERENCES

Thomas R. W., Da Silva L. A. and Mackenzie A. B.,

Cognitive Networks, IEEE New Frontiers in Dynamic

Spectrum Access Networks, DySPAN 2005: 352-360.

Fortuna Carolina, Mohorcic Mihael, Trends in the

development of communication networks: Cognitive

networks, Computer Networks, 2009, 53 (9):

1354-1376.

Ahmad Belhoul, Y. Ahmet Şekercioğlu, Nallasamy Mani,

Mobility-Aware RSVP: A framework for improving

the performance of multimedia services over wireless

IP-based mobile networks, Computer

Communications, 2009, 32 (4): 569-582.

Spiridon Bakiras, Victor O. K. Li. A scalable architecture

for end-to-end QoS provisioning, Computer

Communications, 2004, 27 (13): 1330-1340.

Ping P. Pan, Ellen L. Hahne, Henning Schulzrinne. BGRP:

Sink-Tree-Based Aggregation for Inter-Domain

Reservations [J], IEEE Journal of Communications

and Networks, 2000, 2 (4): 157-167.

RSVP-CN: ONE IMPROVED RSVP MECHANISM ADAPTED TO COGNITIVE NETWORKS

261

Mohit Chamania, Y. Xiaomin Chen, Admela Jukan, Franz

Rambach, Marco Hoffmann, An adaptive inter-domain

PCE framework to improve resource utilization and

reduce inter-domain signaling, Optical Switching and

Networking, 2009, 6 (4): 259-267.

R. Bless. Towards scalable management of QoS-based

end-to-end services[C]. Proc. Network Operation and

Managment Symposium, 2004, 1 (4):293-306.

Bongkyo Moon, Hamid Aghvami. QoS mechanisms in

all-IP wireless access networks [J]. IEEE Journal on

Selected Areas in Communications, 2004, 22 (5):

873-888.

Jiongkuan Hou, Yuguang Fang. Mobility-based channel

reservation scheme for wireless mobile networks[C].

IEEE, Wireless Communications and Networking

Conference: IEEE, 2000: 527 – 531.

Pu Jing. Task dividable based reservation for grid

computing, Computer Engineering and Applications,

2008, 44 (12): 118-120. (In Chinese)

Guo Meng, Zhang Yong-jun, Peng Yu-Xing. A

Pre-Reservation Technology Based on RSVP,

Application Research of Computers, 2004, 21(2):

1001-3695. (In Chinese)

ICISO 2010 - International Conference on Informatics and Semiotics in Organisations

262