Bell Laboratories
Lucent Technologies
Mobile Networking Research Department
101 Crawfords Corner Road
Holmdel, New Jersey 07733
Room 4F515
thuel@lucent.com
|
Sandra R. Thuel
Bell Laboratories Lucent Technologies Mobile Networking Research Department 101 Crawfords Corner Road Holmdel, New Jersey 07733 Room 4F515 thuel@lucent.com |
AQUA: Assured Quality Routing for Inter-Provider Internet Services
The suitability of the Internet to support quality-competitive telephony services has long been touted as a key challenge to the widespread deployment of VoIP. To enhance the Internet’s inherent best-effort transport mechanisms, many on-going efforts are defining and standardizing both IP routed and MPLS-based traffic engineering (TE) solutions for QoS transport within a single provider’s administrative domain. However, inter-domain transport remains best-effort as traffic is routed with BGP4, which was not designed to consider performance objectives in the selection of inter-domain paths. Though BGP policies could be tuned to favor good-quality paths, today’s economics of public peering encourage providers to minimize the distance that traffic travels in their networks, resulting in the infamous hot-potato routing problem. Under these circumstances, emerging service providers seeking inter-domain quality assurances are turning to private peering, multi-homing, and BGP route optimization technology to monitor and inject good-quality routes into the network. Instead of influencing the BGP-based native IP routing fabric to pick good quality routes for all traffic, we are investigating dynamic routing approaches to seek and maintain good inter-domain routes for quality-sensitive traffic, such as Voice over IP (VoIP). These approaches fundamentally seek to find and exploit inter-domain path diversity while also providing quality assurances through resource reservations, at time and capacity granularities commensurate with current intra-domain traffic engineering practices.
Packet Radio Access Networks
Wireless carriers are increasingly realizing the benefits of evolving their cellular radio access networks (RANs) from embedded circuit-based access architectures to networks that support packet transport, particularly IP traffic. A carefully designed converged access network facilitates support for future wireless data services while significantly reducing the costs of deploying and maintaining separate voice and data access networks. To enable this vision, we conducted several research projects, targetting the following key areas:
- Wireless and Mobility-enhanced Routers: Cellular access networks have certain radio control and traffic processing functions that are typically performed by specialized equipment designed to flexibly handle the needs of wireless carriers. In an effort to provide cost- and performance-efficient converged solutions, we investigated the viability of equipping commercial-grade IP routers with wireless support while maintaining the call capacity and voice quality of today’s cellular access networks. The design and development of RIMA (Router for Integrated Mobile Access) was our first leading-edge wireless-enabled router, integrating support for the European GPRS radio standard for data onto an IP access router. In addition, RIMA supported data user mobility through Mobile-IP and HAWAII, and served as an access interface for GSM voice traffic. The success of this effort led to the development of CWAVE, a CDMA Wireless Access Voice-Enhanced router. CWAVE is an access IP router capable of managing the 3G CDMA voice traffic of a large number of cellular base stations with an efficient IP transport of voice frames over the access network.
- Efficient architectures for Radio Network Controllers (RNCs): Radio access networks based on the UMTS standard rely on RNCs as the central element for most of the wireless traffic processing and control of radio resources. As a result, it is critical that the RNC platform supports a broad range of wireless carrier needs in a scalable, flexible, and reliable manner. We investigated and proposed architectural enhancements to Lucent's RNC products for intelligent load balancing amongst platform resources, to improve its resilience to changes in traffic mix, in user density (e.g., congestion hot-spots), and in mobility patterns while maintaining high utilization efficiency. These patent-pending enhancements provided the flexibility needed to gracefully support RNC growth as hardware or software resources are added to match increased capacity demands and to handle configuration changes due to failures.
- Quality of Service (QoS) via congestion control: 3G RANs are slowly evolving from traditional voice-centric, point-to-point backhaul links between base stations and Radio Network Controllers (RNC’s) to packet-based backhaul networks carrying both voice and data over IP. A key technical challenge lies in defining appropriate backhaul provisioning and congestion management strategies that can scalably accommodate increasing, bursty data traffic without compromising voice traffic quality nor requiring costly over-provisioned links. This scalability and backhaul utilization efficiency can be met by judiciously deploying QoS support in the RAN, which offers the additional advantage of supporting future value-added services that rely on traffic differentiation. Our research introduced several novel congestion control strategies for managing packet loss in a RAN, which controlled the amount of traffic in the RAN by adjusting call admission capacity at the cells and intelligently dropping certain redundant legs of active users in CDMA soft-handoff state. In addition, we investigated the use of active queue management (AQM) support in the RAN, more conventionally used in wireline data networks. Among other findings, results suggest that these controls can yield up to a 30% reduction in the size of backhaul facilities without compromising voice traffic quality.
HAWAII Microbility management:
While the standard Mobile-IP (M-IP) protocol for mobility management in IP networks is adequate for portable hosts, wireless, mobile hosts are susceptible to large handoff delays and packet loss when moving across wireless base stations. We developed an efficient micromobility protocol called HAWAII that addresses this problem by providing localized mobility management on the wireless access network while relying on Mobile-IP solely for handling mobility across IP access domains. Our approach maintains connectivity to a mobile host by installing host-based routing entries on selective routers in the access domain as the host moves, instead of changing the IP address of the mobile as is done in M-IP. This improves handoff latency by significantly reducing the frequency of costly M-IP registrations across the network, while facilitating the support of conventional wireline IP QoS protocols on wireless hosts. We implemented and evaluated HAWAII on a network testbed.. Another related wireless access problem addressed in this project was that of supporting dynamic home addressing for mobile hosts that power up away from their home networks. These remote hosts cannot rely on DHCP's conventional use of broadcasting to discover a home addressing server. However, our proposed Transient Tunneling Protocol, introduced a solution that requires no changes to protocol standards and minor changes to servers.
These are some areas I've worked on in the past:
Sandra Thuel / thuel@lucent.com
To send comments or suggestions concerning this page.