Research Challenges and Mission

Future multiservice (multimedia) networks will carry traffic such as video, audio and computer data with a diverse range of quality-of-service (QOS) requirements. In addition, they are expected to have a very large number of physical and logical entities, services and users to be managed and controlled. We believe that the development of open programmable multimedia networks is one of the key research challenges that faces the networking community as we move towards the new millennium. To address this challenge the COMET Group has initiated a number of new projects and international forums to promote the ideas which drive our research. In what follows we describe some of these new initiatives, forums and research projects.

Key Initiatives in ATM, Internet and Mobile Multimedia Networking

The COMET Group has played an important role in establishing new international forums to promote and address the need for research into open architectures, open signalling and network programming for ATM, Internet and mobile networks.

ATM Initiatives The group's research in ATM-based broadband networks has considerably expanded beyond investigating network architectures, and network control and management, into the area of open signalling and telemedia. The work in open signalling was prompted by the simple observation that the origin of the prevalent thinking in signalling and service creation dates back to the late sixties and is based on the fundamental assumption that the ``intelligence'' required for service creation only resides inside the network. We are convinced that this assumption is now outdated and that a fresh approach to the subject is required.

In response to this challenge the COMET Group played an important role in establishing an international working group called OPENSIG (open signalling) for actively exploring network programmability and next generation open signalling technology for service creation and deployment on ATM-, Internet- and mobile multimedia networking platforms. In that spirit the COMET Group aims to make the process of service creation, deployment, control and management of new network services as easy as building and launching new applications on a PC. The OPENSIG group meets on an informal basis every six months to exchange ideas and discuss appropriate forms of collaboration with the stated goal of promoting research and development on open signalling and service creation.

As a result of the tremendous success of OPENSIG, the IEEE Communications Society is sponsoring a new conference on Open Architectures and Network Programming (OPENARCH '98). The first OPENARCH conference will be held on April 3-4, 1998 in San Francisco and will be co-located and organized in conjunction with INFOCOM'98. OPENARCH will offer a forum for the communication of experimental as well as theoretical results aimed at a better understanding of the overall networking architecture and its realization in software. It will encourage a shift of the processes of service creation, resource allocation and control from ad-hoc solutions to a discipline of network programming.

IETF Initiatives The Real-Time Transport Protocol (RTP), developed by the Audio/Video Transport (AVT) working group within the IETF with the help of H. Schulzrinne, is gaining widespread acceptance as the common transport protocol for the delivery of continuous-media information over the Internet, including Internet conferencing, telephony and media-on-demand. Its adoption by the ITU-T as part of its H.323, a first for an Internet standard, ensures its use for Internet telephony and LAN-based conferencing in general. RTP describes a data packet format as well as a control protocol (RTCP) that allows participants to gauge the group size, to establish the identity of other group members and measure the quality-of-service received by members. The rate of the control protocol is self-regulated to a small fraction of the data bandwidth.

RTP is currently a Proposed Standard, soon to be ``upgraded'' to a Draft Standard. One of the problems to be addressed during this transition is the use of RTP in very large multicast groups, as might be found for CATV-like distribution of audio and video programming. In that scenario, with potential audiences of thousands to several million, the self-regulating feedback mechanism and group size estimation of RTCP ceases to function properly. Also, if a large number of users join a multicast group simultaneously, a large surge of control traffic could overwhelm parts of the network. We have developed and analyzed a number of remedies for these problems. These algorithms ensure that the rate of feedback scales regardless of group size and join timing as well as allow quick audience size estimation. They have been submitted to the AVT group for consideration and will likely be integrated into the next release of the standard.

While RTP provides the common base for data transport, the Internet currently does not have a well-developed set of application-layer signaling protocols for Internet conferencing, telephony and media-on-demand services. We have contributed to two protocols to address both the initiation of conferences and the control of streaming audio and video sessions. Both protocols are loosely based upon HTTP, allowing to reuse both protocol features such as encryption, authentication, billing negotiation and content labeling as well as their server implementations.

The session initiation protocol (SIP) offers the standard telephony features, including call forwarding, support for automatic call distribution, and call transfer, allowing the replacement of both traditional PBXs and telephone switches with simple SIP gateways. SIP is independent of the underlying transport protocol and can initiate RTP sessions as well as other data streams, unicast or multicast. Thus, it could also be used for ATM-based conferencing and telephony. The address structure for SIP is designed to allow gatewaying into the existing telephone network. SIP offers rich functionality, yet is significantly simpler than H.323.

Currently, multimedia streaming services such as VDO or RealAudio use their own simple streaming protocol. In collaboration with Progressive Networks (RealAudio) and Netscape, we are designing a stream control protocol for streaming multimedia services, such as audio-on-demand or video-on-demand. The Real-Time Stream control Protocol (RTSP) supports multimedia sessions originating at different servers, recording, device control and the integration of streaming multimedia into conferences initiated by SIP. Both SIP and RTSP are currently under consideration within the IETF MMUSIC conference control working group.

Mobile Initiatives Recently there has been considerable interest in the emergence of wireless ATM (WATM) as a possible candidate technology for next generation mobile multimedia communication systems. As multimedia applications migrate to mobile devices, wireless extensions to broadband networks are required to support the seamless delivery of voice, video and data with high quality. In this context WATM is intended to be a direct extension of the existing ATM network with uniformity of end-to-end QOS guarantees.

In early 1996 the ATM Forum established the WATM working group to address the challenge of providing broadband services over an air-interface. The primary goal of the WATM working group is to develop a set of specifications for location management, ATM mobility and a new 25 Mbps air-interface.

The COMET Group has been actively involved in shaping the discussion within the WATM working group and in the broader mobile and wireless networking community at large. First, we have participated in the WATM working group since its creation. Our involvement in this new initiative is to track the development of new standards, to exploit the emergence of new techniques through the development of software solutions and, to feed our research results into the working group. Second, prior to the establishment of the ATM Forum WATM working group we organized the First Workshop on Wireless ATM Networking. The workshop was a great success bring together leading researchers involved the implementation of prototype WATM networks and the key people who went on to establish WATM working group.

Benefits of COMET Research

The investigations of the COMET Group have been regularly published as papers in leading journals and conferences. A complete list is given in the Publications section.

In addition we made our software prototypes also available through the WWW. Currently the following software can be downloaded:

• kStack: User Space Native-Mode ATM Transport Layer with QOS Support
• RTP: A Transport Protocol for Real-Time Applications
• TREX: Telecommunications Research Exchange: A Dynamic Auctioning Game for Spectrum Allocation in Wireless Networks.
• xbind: A broadband kernel for devising the next generation technology of service creation, deployment and management on ATM based broadband networks.

Research Focus

In this section we highlight a number of new COMET research initiatives in ATM, Internet and mobile multimedia networks. Following this we provide an overview of each on-going project in the form of a bullet report.

Broad access to open multimedia networks will play an important role in enabling the emergence of new broadband networking technologies and programming paradigms. A major thrust of the COMET group is the development of an open programming environment that facilitates the easy creation of distributed multimedia services on top of ATM, Internet and mobile networks as illustrated in Figure 1. By open, we mean that the proposed platform must support a set of APIs for resource control and management which third party service providers can use when developing new services. By programmable, we mean that these APIs should be `high-level' enough to allow the service specification and creation process to be carried out via a high-level programming language utilising distributed systems technology such as CORBA.

Figure 1: Model of programming for multimedia networks

Since the Fall of 1994 we have been experimenting with a broadband kernel called xbind. The broadband kernel is as a programmable operating platform that supports the creation, deployment and management of networked multimedia services (e.g., connection management, route management, admission control, QOS mapping, etc.) and mechanisms for efficient resource allocation. The term `kernel' is deliberately used to draw a parallel between its role as a resource allocator and extended machine, and that of a typical operating system. The broadband kernel behaves as a resource allocator because it mediates and arbitrates between conflicting requests for resources made by various parties in the system. It functions like an extended machine because it provides a simplified means of accessing fundamental system services by abstracting away the operational complexities of provisioning these services.

Resource allocation in xbind is based on game theoretic models. By taking an economic (market based and game theoretic) approach in the engineering of multimedia networks, we seek solutions where the intelligence and decision making is distributed and thus scalable, and the objective of a more efficient and fair utilization of shared resources results from the induced market dynamics. Thus, we are borrowing, and adapting the tools of game theory from Economics and using them to solve problems of resource allocation (such as admission control, routing and flow control) in multiservice networks. By viewing a network as a collection of resources which users are selfishly competing for, this approach gives rise to efficient, decentralized algorithms, and leads to network architectures which provide explicit QOS guarantees.

The players in the network economy are software agents, rather than humans. Agents acquire resources, such as bandwidth and buffer space, from the network on behalf of applications (video, voice, data transfer). Under appropriate rules of interaction, the collective actions of all the agents constitute a distributed intelligence, superior to that of any single controller. Thus the challenges are to analyze noncooperative behaviour and algorithmic strategies, and to design the mechanisms (rules of the games) that will ensure the desired outcomes.

Advanced Multimedia Internet Services

From modest experiments transmitting Internet Engineering Task Force meetings in 1992, the Internet multicast backbone and the associated set of IP multicast tools have seen widespread use. Research into packet audio and video in the Internet, however, go back to the early 1970s. Based on these experiences, the Real-Time Transport Protocol (RTP) was developed, which is now on the Internet standards track. A number of research organizations and companies have developed interoperable, RTP-based audio and video tools. Also, as an Internet first, the ITU has adopted RTP as a part of their multimedia conferencing architecture for LANs, commonly known as H.323. A number of well-known vendors have started shipping H.323-based Internet telephony products.

The demand for Internet multimedia teleconferencing grows dramatically by the day, but much work needs to be done before truly scalable transport and signalling protocols, and teleconferencing tools are in place. The COMET Group is actively meeting these challenges [3]. Future work in the group includes the development of a Session Initiation Protocol (SIP) that performs all standard telephony signalling functions, including automatic and manual call forwarding. In addition, new feedback modes for RTCP need to be developed to address some of the scaling problems that manifest as the number of participants in an RTP session scales from hundreds to potentially thousands. Other multimedia Internet services research includes the development of a Real-Time Stream control Protocol (RTSP) for controlling the retrieval of continuous media over the network.

Currently, we are developing a suite of protocols and application building blocks which address these open research issues. Our goal is to provide scalable solutions for current and future multimedia services as the demand for a multimedia-ready Internet accelerates.

Wireless Media Systems

The development of next generation mobile multimedia communications systems presents a number of technical challenges which are thus far unresolved. These challenges, mainly due to large-scale mobility requirements, limited radio resources and fluctuating network conditions, fundamentally impact our ability to deliver multimedia flows over mobile and QOS fluctuating networks.

Figure 2: Mobiware: QOS-aware mobile middleware

First, it is essential that QOS assurances be given for the transfer of audio and video flows to mobile devices as they roam between cells in cellular systems. Second, future mobile communications systems must be able to provide dynamic re-routing of a set of multimedia flows associated with a mobile device from one base station to another in a timely manner, without significantly interrupting the flows in progress and with a smooth change in the delivered quality. Third, existing multimedia transport systems are ineffective when operating in environments where widespread mobility and changing network characteristics are dominant. New mobile middleware needs to be designed that can operate in the face of QOS-varying channels and rapid device mobility. These protocols are yet to be designed and implemented.

Although researchers have addressed architecture and algorithms for supporting QOS in the wireline networking environment, little attention has been paid to the impact of mobility on QOS in multimedia communications. To address this research challenge the COMET Group has recently initiated a new research effort in wireless media systems [4] to develop solutions for QOS controlled mobile multimedia networks. The research is directed toward the development of mobile middleware for ATM based broadband networks (as illustrated in Figure 2) and the next generation mobile integrated services Internet.

Prototyping Multimedia Networks

In order to study multimedia networks under various scenarios, such as different load patterns, network sizes and management operations we have built a platform for developing and evaluating control systems for emerging multimedia networks. The platform allows us to closely approximate the functional and dynamic behavior of network control systems and provides support for real-time visualization and interactive emulation, . The current implementation runs on the IBM SP2 parallel processor at the Cornell Theory Center, that is connected to a graphics workstation in our laboratory at Columbia University via an ATM link. The platform is currently used in various projects in the COMET group that are all aimed at developing and evaluating network architectures. The approach taken is real-time control based and as such differs from the current work of the Telecommunications Information Networking Architecture Consortium (TINA-C) which is essentially management based with a focus on data and functional modelling.

Figure 3: Prototyping multimedia networks

References

[1] The COMET Group

[2] Open Programmable Multimedia Networking

[3] Multimedia Internet Services Research

[4] Wireless Media Systems Research

[5] OPENSIG

[6] OPENARCH

For more detailed information about our research please consult our most recent Activity Report. For your convinience an up to date overview is given below.

Information Transport	Network Architectures	Network Control
Network and Service Management	Resource Allocation and Networking Games	Telemedia
	Wireless Media Systems

kStack: A User Space ATM Transport Layer with QOS Support

qStack: A Native ATM Transport with QOS Guarantees.

QOS Mapping

Real-Time Traffic Generation and Monitoring

We use this system to to study the behaviour of multiple streams that share and compete network resources. In particular, we are interested in the impact of control actions taken by one injector upon the other. As another application of our system, we measure the network quality of service of distributed multimedia applications that run on our ATM LAN.

The system is also used to generate traffic, including call arrivals and departures, for testing the schedulable region estimation algorithm.

Broadband Virtual Private Networks

Real-Time Emulation and Visualization of Networks on a Supercomputer

We have developed such a platform, which includes an emulator that runs on a parallel machine at the Cornell Theory Center (CTC) and a manager station at Columbia University, which is implemented on an SGI Indigo2 workstation, running a 3D user interface for real-time visualization and interactive control. Both machines communicate through NYNET, an ATM network that connects several research laboratories in New York State.

xbind: A Broadband Kernel for Multimedia Networks

Evaluating the Schedulable Region in Real-Time

In the present project, we concentrate on evaluating the size of the schedulable region in real-time. Due to the randomness of the calls traversing any given link, the schedulable region of that link may vary over time. We have designed algorithms which run on switches for measuring the activity in the output ports. This information is used to determine the size of the schedulable regions of the output links.

Generic Control APIs

Switch Control API with support for QOS

Managing Real-Time Services on Broadband and Multimedia Networks

Network Fault Management

Building Manageable Multimedia Network Services

The aim of this project is to design a new service management model in the context of Enterprise ATM Networks in which telecommunication services are deployed and managed. We consider four types of services in our design in the first instance: virtual connections, virtual paths, virtual networks, and multicast services. In this context the service management activity is split between affecting the control and state of the system. The initial focus of the work is on configuration and performance management of existing service instances. In this regard the service management activity becomes similar in context to the QOS management activity from a provider viewpoint. The proposed service model is simplified to handle a limited set of entities interacting in a hierarchical structure.The management Application Programming Interfaces (APIs) are defined as object interfaces based on the Common Object Request Broker Architecture (CORBA) distributed platform.

Resource Allocation and Networking Games

Thus, we are borrowing, and adapting the tools of game theory from Economics and using them to solve problems of resource allocation (such as admission control, routing and flow control) in telecommunication networks. By viewing a network as a collection of resources which users are selfishly competing for, this approach should give rise to efficient, decentralized algorithms, and lead to network architectures which provide explicit Quality of Service guarantees, the crucial issue in highspeed multimedia networks. Broadly, we call this approach Networking Games.

The rationale for these approaches is that centralized control is not possible in reality because of the diverse ownership of the network resources. More fundamentally, unlike for example a computer where all resources are ruled by the operating system, a large multimedia network with QoS would be too complicated and far-flung to be under a single authority, or even to define centralized control objectives. Thus the assumption of selfish users is not due to pessimism, but rather a recognition that technology and nature make it impossible to be unselfish, or even to know what it is to be unselfish. Finally, with the Networking Games approaches, the crucial issue of pricing can be resolved from the ground-up in the engineering of the network, rather than having, as in existing communication networks, a necessarily arbitrary price structure imposed post-facto.

The players in the network economy are software agents, rather than humans. Agents acquire resources, such as bandwidth and buffer space, from the network on behalf of applications (video, voice, data transfer). Under appropriate rules of interaction, the collective actions of all the agents constitute a distributed intelligence, superior to that of any single controller. Thus the challenges are to analyze non-cooperative behaviour and algorithmic strategies, and to design the mechanisms (rules of the games) that will ensure the desired outcomes.

Resource Allocation and Networking Games focal areas:

Resource Allocation and Networking Games API

---

Telemedia

Virtual Workshop

As part of the project ATM Management and Control API (F30602-95-R-0143) with Rome Labs, we plan to design and implement a Virtual Workshop service. The Virtual Workshop service is a distributed multimedia application that provides various forms of multimedia associations among VW participants. Like a "real workshop" a Virtual Workshop features interactions such as sessions and tutorials, as well as one-to-one conversations among participants. In contrast to a "real workshop" however, the VW participants are physically distributed instead of meeting in the same physical location. They make use of the broadband communication infrastructure and the VW service in order to interact.

The Virtual Workshop is an advanced multimedia service, that requires interaction and cooperation among various entities : the Virtual Workshop application, the network control system, the service management system, and the network management system -- and thus readily allows us to experiment with our interface designs under demanding service requirements. We use xbind as a platform for implementing the generic interfaces described in the previous sections of this project. The xbind system already provides us with the fundamental underlying support services necessary in a multimedia network with QOS guarantees.

Service Modeling

The recent move towards market deregulation and open competition has sparked a wave of serious introspection in the telecommunications service industry. Telecom providers and operators are now required to open up their primary revenue channels to competing industries. In this research, we explore the general notion of communication services and propose a new service architecture which better reflects the needs and structures of the future telecommunications service market. Our model is heavily API-oriented and extends previous work on multimedia network broadband kernel services.

---

	Activities	Courses	Launch Pads	Location	Members
	Publications	Research	Software	Sponsors	xbind

Have a question or comment for the COMET Group? Please email us at [email protected]!

---

Please email your suggestions, comments, and problems only about this website to [email protected].

Last Update: Sat Oct 5, 1996.