QOS-AWARE POLICY BASED ROUTING FOR MESH

NETWORK ENVIRONMENTS

Mathias Kretschmer, Ilka Miloucheva

Fraunhofer Institute, Schloss Birlinghoven, Germany

Dirk Hetzer

T-Systems, Berlin, Germany

Keywords: Multi-path routing, QoS, policy, resource planning, wireless mesh network, broadcast media, mesh gateway.

Abstract: The design of a policy oriented configurable multi-path routing architecture for enhanced mesh network

infrastructures is discussed. The goal is optimal routing path selection between mesh clients and mesh router

gateways dependent on application classes. Policies are used to dynamically configure the route path

selection for QoS-aware applications taking into account resource reservation requests for application traffic

(in advance reservation, on-demand resource allocation, etc) and business goals of the policy actors, i.e.

end-users, service providers and network operators. Interactions of the mesh routing protocol facilities with

the policy management and resource planning components for advance and on-demand resource allocation

are considered. Scenarios are aimed at supporting QoS-aware applications (such as IPTV, VoD, GRID,

VoIP, real-time, mission critical, content delivery on-demand, software and large file downloads) using a

wireless mesh network backbone infrastructure (based on IEEE 802.11, IEEE 802.16) enhanced with

broadcast technologies (DVB-T, DVB-H).

1 INTRODUCTION

Wireless mesh networks are promising solution for a

number of emerging scenarios – broadband home

networking, “city” community, neighbourhood,

transportation, health and medical networking

systems (Akyildiz et al., 2005). In the new business

models for wireless mesh network infrastructure, it

is a challenge to support a wide range of QoS-aware

applications with specific resource and delivery

requirements, such as IPTV, VoD, GRID, VoIP,

real-time, mission critical, content delivery on-

demand, software and file downloads. The interest

of providers and operators is to efficiently use the

wireless mesh infrastructure considering the

dynamic resource requirements of QoS-aware

applications and the optimisation of the routing for

different kind of application traffic.

Resource requests can be focussed on selection

of most the appropriate network and router, resource

reservation in advance, on-demand reservation,

reservation with different level of QoS guarantees

(Hetzer et al., 2006) and others. There is a need for

new management strategies including interactions of

resource management and routing facilities in order

to enhance the QoS guarantee depending on the

specific application resource requests in

heterogeneous wireless mesh environments.

In this paper, a QoS-aware policy routing

architecture is proposed, which is aimed at flexible

support of QoS-aware Internet applications in

heterogeneous networking environments dependent

on the policies of end-users, service providers and

network operators. The particular focus is the design

of components and their interactions for QoS-aware

policy routing in wireless mesh infrastructures

(IEEE 802.11, IEEE 802.16) enhanced with

broadcast technologies (DVB-T, DVB-H).

With the synergy of the broadcast and Internet

worlds, there is an increasing demand for routing

facilities considering the specifics of broadcast

networks and their cost-efficient usage for specific

kind of applications (Near-Video-on-Demand,

IPTV, content delivery). In order to support efficient

resource usage and enhanced QoS, policy based

multi-path routing facilities for QoS-aware

applications are proposed, which use resource

305

Kretschmer M., Miloucheva I. and Hetzer D. (2007).

QOS-AWARE POLICY BASED ROUTING FOR MESH NETWORK ENVIRONMENTS.

In Proceedings of the Second International Conference on Wireless Information Networks and Systems, pages 289-294

DOI: 10.5220/0002151602890294

 SciTePress

(bandwidth) planning and QoS/SLA monitoring

information collected at the mesh router gateways.

This paper is organised in the following sections.

Section 2 evaluates QoS issues of current routing

protocols for wireless mesh network infrastructure.

Section 3 discusses components for QoS-aware

policy routing. Section 4 describes a scenario based

on a wireless mesh infrastructure enhanced with

broadcast media. In section 5, a policy management

interface for flexible QoS routing configuration is

discussed. Section 6 concludes this paper.

2 POLICY ROUTING FOR

WIRELESS MESH NETWORKS

The routing in wireless mesh networks is primarily

based on the concepts of routing protocols for ad-

hoc infrastructures. In order to support QoS-aware

routing of value added services and applications

with specific QoS requests, the routing protocols can

be enhanced with policy based routing strategies.

2.1 Routing Protocols

IP has been accepted as network layer protocol for

integration of wireless mesh networks.

Wireless mesh networks can be deployed as

infrastructure/backbone and client architectures

connected to the Internet. The connectivity of the

mesh infrastructure / backbone to the Internet core is

based on mesh gateways. Mesh routers are forming a

multi-path routing infrastructure for mesh clients to

the mesh gateways (Akyildiz et al., 2005) (figure 1).

Actual requirements for routing in wireless mesh

networks are support of alternative (multiple)

routing paths, efficient resource usage, support of

QoS routing in heterogeneous environments, as well

as enhanced routing reliability.

For the interconnection and routing in mesh network

environments, different protocols developed for

mobile ad hoc networks can be efficiently applied

(Conti et al., 2007). Such protocols are for instance:

- On-demand Distance Vector Routing (AODV)

(Perkins et al., 2003).

- Topology dissemination based on reverse path

forwarding (TBRPF) (Ogier et al., 2004),

- Dynamic Source Routing (DSR) [6],

IP core

WLAN

Mesh gateway

Wireless

mesh

infrastructure

Figure 1: Wireless mesh infrastructure.

Algorithms and protocols for mesh routing can be

classified depending on their mechanisms and

applications into:

- On-demand routing protocols, which create a

route between a pair of source and destination

nodes, when the source node actually needs to

send packets to the destination (see, DSR

(Johnson et al., 2004);

- Proactive Routing Protocols, where each node

maintains one or more tables containing routing

information to every other node in the network.

According the method for forwarding packets

along routes, the proactive routing differentiates

between source routing (for instance, Link

Quality Source Routing – LQSR (Draves et al.,

2004)) and hop-by-hop proactive routing.

Intelligent mechanisms for dynamic routing

optimisation in mesh network infrastructures can be

integrated in order to analyse routing stability,

improve performance and load balancing, as well as

to support autonomic design considerations.

ROMER (Resilient Opportunistic Mesh Routing

Protocol), for example, is a routing protocol

balancing between long term route stability and

short term opportunistic performance (Yuan, et al.,

2005).

Dependent on application and business goals, the

efficiency of the routing algorithms can be evaluated

in terms of performance metrics (Yang, et al., 2005).

Such metrics are blocking probability by attempting

several routes in parallel, protocol overhead for

selection of specific routes (routes with maximum

QoS guarantee and with minimal cost), route

recovery time in case of failure.

WINSYS 2007 - International Conference on Wireless Information Networks and Systems

306

2.2 QoS-aware Policy Routing

The design of QoS-aware policy routing depends on

the application (business goals and scenarios), as

well as on the specific structure of the infrastructure.

Policy oriented routing mechanisms for Internet

protocols are discussed in RFC 1102 (Clark et al.,

1989). To support specific business goals and

policies, the design of QoS-aware routing for mesh

networks can be based on application and overlay

routing techniques (see (Shi et al., 2002), (Kwon et

al., 2002), (Zhang, et al., 2005)).

Overlay routing protocols can be designed for

different topologies – full mesh, minimum spanning

tree, mesh-tree, adjacent connections (Li et al.,

2004). Algorithms for multi-path routing combined

with reserving resources in parallel along several

routes connecting end-users are discussed in (Cidon,

et al. 1998).Resilient Overlay Network (RON)

(Andersen, et al., 2001) is an example for

sophisticated policy routing considering source

traffic type and application class. The policy routing

is separated into two components – classification

and routing table formation. The policies allow the

detection and recovery from path outages and

periods of decreased performance. Further related

work is aimed at scalable Distributed Object

Location and Routing overlay services exploiting

existing network redundancy by dynamic switching

of traffic onto pre-computed alternative routes

(Ratnasamy et al., 2001), (Rowstron et al., 2001).

The QoS-aware policy routing design proposed in

this paper is based on the overlay technique and

implies interaction of resource planning and routing

facilities.

3 POLICY BASED ROUTING

ARCHITECTURE

The proposed multi-path QoS routing facilities are

based on dynamic configuration and adaptation of

routing tables for application traffic using policies of

different actors – end-users, service providers and

network operators. The policy framework is

designed to consider different resource reservation

strategies for applications, such as:

- On-demand and planned reservation,

- Selection of networks and routers dependent on

application criteria,

- Bandwidth planning and monitoring information

at the mesh gateways.

3.1 Policy Definition

The policy model for QoS-aware routing is based on

IETF policy framework, especially RFC 3060

(Moore, et al., 2001), RFC 3640 (Moore et al.,

2003), RFC 3644 (Snir, et al., 2003).

A policy consists of a set of rules defined by the

policy actor. The rules are specified for different

application classes using condition and actions.

A policy rule P for the QoS-aware routing is set by:

:= C

(App X Act X Res) Æ A (G X Rt X B)

, where

• C is a condition defining resource reservation

options for application classes and policy actors:

• App

is the set of specified application classes;

• Act

describes the profiles of the policy actor;

• Res

is the set of possible resource reservation

strategies;

• A is action updating the routing and resource

planning tables at the mesh gateways;

• Rt

is the set of routing tables;

• G is graph describing the topology of the

wireless mesh infrastructure;

• B is the set of bandwidth plans for the mesh

gateways.

The different reservation strategies depend on the

specific traffic classes, such as content delivery, file

downloads, media streaming (IPTV and VoD),

GRID, real time mission critical and voice over IP.

Examples for different kind of resource reservation

strategies, which can be used in the policy

specification for the QoS-aware policy routing, are

given in figure 2:

Figure 2: Resource reservation strategies.

3.2 Functional Design

The architecture is based on existence of alternative

routes between mesh clients and gateways of a

wireless mesh infrastructure / backbone (see, figure

1). The configurable routing path selection is using

Advance

resource

reservation

On-demand

reservation

Preferences

for routing

path & mesh

Future

reservation

(starting at

given time,

within

specified time

interval ,

periodically)

Immediate

reservation for

a given time

interval

dependent on

application

QoS and

demands

- Costs

- Reliability

- QoS guarantee

- Availability

- High

Bandwidth.

QOS-AWARE POLICY BASED ROUTING FOR MESH NETWORK ENVIRONMENTS

307

the specified policies of the different actors in order

to select dynamically the optimal routing path and

mesh gateway for the particular application.

Based on a common policy repository, different

components are interacting:

- Policy management system enforcing the

facilities for route table configuration and

bandwidth planning / monitoring;

- Facilities for automated routing table

configuration at the routers of the mesh

infrastructure;

- Toolkits for QoS/SLA monitoring and bandwidth

planning at the mesh gateways.

Interfaces between these components allow the

dynamic reconfiguration of the routing information

for the application classes and the adaptation of the

bandwidth planning information at the mesh

gateways based on the specified policies.

The functional design and the interactions of the

components are shown in figure 3.

Using management interfaces, the different actors

enter the policies into the system.

The policies are checked for consistency, translated

and stored in policy repositories.

Because the policies are set by different actors, their

parameters (specified by conditions and actions) can

be automatically changed by the Policy Decision &

Adaptation Manager according the hierarchical

relationships of the actors (end-users, service

providers and network operators) and their

QoS/SLA for the application classes. Furthermore,

based on the feedback of the bandwidth reservation

planning and QoS/SLA monitoring data at the mesh

gateways, the policies can be dynamically adapted

and optimised by the policy decision manager.

In order to trigger the final policies, the policy

enforcement manager interacts with the routing and

bandwidth planning components. The results can be:

- Dynamic configuration of the routing tables of

the infrastructure for application classes based

on the specified policies;

- Bandwidth plan updates at the mesh gateways

according the selected routing paths.

etwork

operator

policies

Service

provider

policies

Customer

(end-user)

policies

Consistency check and polic translation into

Internal presentation

Policy repository

Policy decision

and adaptation

Policy

enforcement

Routing

configuration

Bandwidth

planning

configuration

Heterogeneous

Wireless mesh and

broadcast

infrastructure

Figure 3: Functional design of QoS policy routing.

For the bandwidth planning and QoS/SLA

monitoring, the QORE system for advance resource

reservation can be used, which supports the

bandwidth planning for different kinds of

application classes and scenarios at the routers of

networking infrastructures (see (Hetzer et al. 2006),

Miloucheva et al. 2007)).

4 SCENARIO

A simple scenario for multiple path routing aimed at

providing enhanced QoS-aware policy routing for

multimedia content delivery of streaming content

(Video-on-Demand, TV channels, news,

advertisement, software downloads) is shown. The

scenario is based on infrastructure / backbone

wireless mesh networks (WLAN IEEE 802.11,

WIMAX IEEE 802.16) enhanced with broadcast

media (DVB-T, DVB-H) connected to the core

Internet (see figure 4). Using policies, the users can

optimise the routing path for the multimedia content

delivery dependent on the resource reservation

strategies of the application.

WINSYS 2007 - International Conference on Wireless Information Networks and Systems

308

IP core

WLAN client

UMTS client

DVB-T

gateway

Best effort

gateway (3)

Wireless

mesh

infrastructure

Support of

Advance

reservation (1)

Gateway

Enhanced

QoS (2)

DVB-T

Figure 4: Scenario for QoS-aware policy based routing.

In the scenario, different routing paths are possible

for application classes dependent on the specific

requirements of the users:

- The user specifies “advance reservation”

strategy defining the delivery time for planned

and periodical applications. For instance, fixed

time can be specified for a requested TV

program, multimedia conference or

entertainment / remote teaching applications.

Future time interval (earliest or latest delivery)

can be requested for applications, such as

Video-on-Demand (VoD) and file downloads.

- The immediate reservation of resources is

specified by “on-demand” resource reservation

strategy. This strategy can be used for

unplanned delivery of content and mission

critical applications.

- When a user requests “best effort” service, an

appropriate gateway is selected by the policy

system and assigned for routing of the

application. Minimal QoS guarantees are

provided for the user traffic.

Assuming that for the multicast and multimedia

content delivery applications with “advance resource

reservation” requests, the routing is specified using

the mesh gateways for the DVB-T networks.

The path selection can be dynamically changed for

particular user profile and application class

depending on further policies aimed at “cost

efficiency”, “reliability”, “high QoS provision”.

The benefit is that the routing paths can be selected

and dynamically changed according to the policies

of the different actors (end-users, service providers

and network operators) and their dependencies.

5 POLICY MANAGEMENT

INTERFACE

The routing policies in wireless mesh infrastructures

considering QoS-aware applications are specified

based on parameters and options using appropriate

Graphical User Interfaces (GUIs) for users, service

providers and network operators.

Policy management interfaces allows specification

of policy parameters, check of policy dependencies

considering actors relationships and policy

translation into internal policy presentations.

An example of a GUI for dynamic selection of

appropriate routes for QoS-aware applications using

policy translation is shown in figure 5.

Figure 5: User interface for policy based routing.

The policy is identified by the policy name, policy

actor profile and QoS/SLA. The parameters of this

interface are related to ontology based domain

descriptions, based on which the relationships

between the parameter values are obtained.

The policies are translated into the internal policy

repository representations. The policy repository

functions allow the performing of operations for

entering, changing, deleting of policies, as well as

browsing of policies based on the policy repository.

Policies can be retrieved from the repository and

dynamically adapted using learning algorithms.

Conditions and actions of the routing policies are

specified by a set of parameters describing

- Application/Services, which are using the routing

infrastructure;

- Resource reservation strategy;

- Gateway requirements.

The actions (change of routing table information

and bandwidth plans at the mesh gateways) are

automatically configured by the systems.

QOS-AWARE POLICY BASED ROUTING FOR MESH NETWORK ENVIRONMENTS

309

In the given example, the management interface

allows the selection of the application class (VoIP,

VoD, Mobile TV, File Transfer, Web). The resource

reservation strategy for the particular application is

specified by options for ”Advance resource

reservation”, “On-demand reservation”,

“HighQuality” or “Best effort”. The routing

requirement is given by the “Shortest path” option.

6 CONCLUSIONS

A QoS-aware policy routing architecture based on

the interaction of policy management, routing and

bandwidth planning components for wireless mesh

infrastructures was proposed. The benefit is flexible

configuration of routing tables for routing depending

on application resource requests. This allows cost

efficient routing per application class considering

wireless mesh and broadcast media networks.

Particular advantage of the architecture is dedicated

usage of broadcast media for specific multicast and

multimedia applications. The proposed design is

based on the policy management framework

developed in the EU IST project NETQOS.

ACKNOWLEDGEMENTS

This work was supported by the project NETQOS

funded by the European commission.

REFERENCES

Akyildiz, I.F., Wang, X., Sept. 2005. A Survey on

Wireless Mesh Networks. In IEEE Communications

Magazine.

Hetzer, D., Miloucheva, I., April 2006. Adaptable

bandwidth planning for enhanced QoS support in user-

centric broadband architectures. In World

Telecommunications Congress (WTC). Budapest.

Hetzer, D., Miloucheva, I., Jonas, K., Nov. 2006.

Resource Reservation in Advance for Content On-

demand Services. In Proceedings of Networks

Conference. New Delhi, India.

Conti, M., Giordano, S., April 2007. Multihop Ad Hoc

networking: the theory. In IEEE Commuications

Magaine.

Ogier, R., Templin, F., Lewis, M., Febr. 2004. Topology

dissemination based on reverse path forwarding

(TBRPF). IETF RFC 3684.

Johnson, D. B., Maltz, D. A., Hu, Y.-C., July 2004. The

dynamic source routing protocol for mobile ad hoc

networks (DSR). IETF Internet Draft.

Perkins, C., Beldin-Royer, E., Das, S., July 2003. Ad hoc

on-demand distance vector (AODV) routing. IETF

RFC 3561.

Draves R., Padhye, J., Zill, B., 2004. Routing in

multiradio, multi-hop wireless mesh networks, In

Proceedings of ACM Mobicom.

Yuan, A., Yang, H., Wong, S. H. Y., Lu, S., Arbuagh, W.,

2005. ROMER: Resilient Opportunistic Mesh Routing

for Wireless Mesh Networks, In Proceedings of IEEE

WiMesh.

Yang, Y., Wang, J., Kravets, R., 2005. Designing Routing

Metrics for Mesh Networks. In IEEE Workshop on

Wireless Mesh Networks.

Clark, D., May, 1989. Policy routing in Internet protocol.

In IETF RFC 1102.

IST project. Policy Based Management of Heterogeneous

Networks for Guaranteed QoS. www.ist-netqos.org.

Shi, S. Y., Turner, J., October 2002. Multicast routing and

bandwidth dimensioning in overlay networks. In

IEEE Journal on Selected Areas in Communications

(JSAC), Volume 20, Number 8.

Kwon, M., Fahmy, S., May, 2002. Topology-Aware

Overlay Networks for Group Communication. In

Proceedings of ACM NOSSDAV.

Zhang, X., Zhang, G., 2005. A Multicast Routing

Algorithm for Overlay Network Built on Leased

Lines. In Symposium on Applications and the Internet

(SAINT).

Li, Z., Mohapatra, P., 2004. The Impact of Topology on

Overlay Routing Service. In Proceedings of IEEE

INFOCOM.

Cidon, I., Rom, R., Shavitt, Y., 1998. Multi–path routing

combined with resource reservation, In Proceedings of

IEEE INFOCOM.

Andersen, D., Balakrishnan, H., Kaashoek, F., Morris, R.,

October 2001. Resilient Overlay Networks. In

Proceedings of ACM SOSP.

Andersen, D., Balakrishnan, H., Kaashoek, F., Morris, R.,

2004. The Case of Resilient Overlay Networks. In

Proceedings of IEEE INFOCOM 2004.

Ratnasamy, S., Francis, P., Handley, M., Karp, R.,

Schenker, S., August 2001. A scalable content-

addressable network. In Proceedings of SIGCOMM.

Rowstron, A., Druschel, P., Nov. 2001. Pastry: Scalable

distributed object location and routing for large scale

peer-to-peer systems. In ACM Midleware.

Miloucheva, I., Hetzer, D., Pascotto, R., Jonas, K., January

2007. Resource Reservation in advance for QoS based

Mobile Applications. In International Review on

Computers and Software (IRECOS). Vol. 2, No. 1.

Moore, B., Elleson, E., Strassner, J., Westerinen, A.,

February 2001. Policy Core Information Model-

Version 1 Specification. IETF RFC 3060.

Moore, B., January, 2003. Policy Core Information Model

(PCIM) Extensions. IETF RFC 3460.

Snir, Y., Ramberg, Y., Strassner, J., Cohen, R., Moore, B.,

November, 2003. Policy Quality of Service (QoS)

Information Model. IETF RFC 3644.

WINSYS 2007 - International Conference on Wireless Information Networks and Systems

310