about mhof

Research

As Head of Bell Labs Research, I have been involved in a broad spectrum of exciting leading edge technologies and innovations, ranging from quantum computing, to optical and wireless networking, routing, distributed systems, to software and application technologies.

My core research expertise is in communication support for distributed systems and, more specific, in communication protocols for the Web/Internet and IP-based video/telephony networks. The overall objective is to improve the performance and the scalability of today's Internet and to make the "World Wide Wait" (WWW) into a real interactive experience integrating sophisticated multimedia technology and providing exciting application services. My work extends into cloud computing, television (IPTV, cable TV, Internet TV) and telephony (VoIP and legacy) world, trying to bring the Web/Internet together with more traditional communication infrastructures.

During the early years of my career, I have been working on reliable and scalable multicast and on Quality of Service support for multimedia communication. Later, the work extended to include caching techniques and next generation content networking architectures for the Internet. More recently, I started to look into advanced softswitch architectures, signaling over the Internet and Voice over IP (VoIP) networks. This now extends to video and television services, in particular IPTV, cable TV, Internet TV and the transmission of all this data over future networks.

My work covers design, specification, implementation and evaluation of communication systems and protocols at the network layer and above. The combination of performing original research and affecting the real world is very important to me. My current and past involvements are/were in the following projects:

NetComputing

The "Networked Computing (NetComputing)" project builds a scalable infrastructure for networked computing environments, extending the cloud computing paradigm with coordinated network resources to address interactive/real-time and highly secure services. It allows many services to be deployed and accessed more economically. A NetComputing infrastructure provides both computing resources (data processing and storage) and network resources (data transmission) to service subscribers. Typically, these resources are allocated dynamically in response to user demands and according to current resource status. Our work focuses on creating solutions that provide more efficient utilization of network-connected computing resources, thereby enabling economical and scalable growth of services. The target solutions will replace static, large-scale resource reservation with on-demand, fine-grained assignment of the computing and network resources that are needed by services and applications. Work focus is on techniques for building consistent distributed, shared data storage, and algorithms for dynamic assignment and re-assignment of service tasks to computing and communication resources, as well as the creation of a highly secure networked execution platform.

Multimodal Communications

Multimodal communication increases the richness and range of communication experiences by bringing together multiple means of communication into a single combined session. In such combinations, the weaknesses of one modality are offset by the strengths of another. For example, patient information in an operating room environment may be requested verbally by members of the surgical team (to maintain an antiseptic environment) and then displayed aurally and visually to the surgeons (to maximize their data comprehension). Our techniques are used, for example, to build a multimedia communication system that will allow a remote doctor to use secure high-performance IP-based connectivity to exchange voice, video, and data with onsite surgeons in one or more operating rooms.

Massive Data Dissemination Networks (MadNets)

The "Massive Data Dissemination Networks (MadNets)" project is developing unique infrastructure elements that will make the distribution of massive amounts of data economically viable while maintaining the service quality expected by the consumer. The market for content distribution is exploding. But the shift towards Web 2.0 style communications with emphasis on social networks and user-generated content pushes existing solutions beyond their limits – it significantly lowers the efficiency of approaches such as traditional caching, multicasting or broadcasting. The MadNets project meets these challenges – it overcomes these inefficiencies through invention of new content routing schemes that leverage peer-to-peer technologies for improved network utilization, intelligent content placement algorithms for faster access to content, and prediction-based content push algorithms that reduce the peak loads, thus reducing the infrastructure costs. Bell Labs is basing this work on extensive experience gained from more than a decade of work on content networking and is applying it in collaboration with several Business Units.

User-Aware Communications

Many emerging and envisioned future communication services process and display large amounts of information or information compiled from multiple sources. To help people deal with these collections of data, information selection, processing, delivery and display must be tailored to individual consumers. Our work seeks to create means of personalizing information access, selection, and display in multi-screen environments (TV, laptop, mobile, etc.) as well in the future pervasive environments (with communicating objects).

Digital TV

The Digital TV Blended Services project is developing server and network components for blending digital TV with other communication services. This project is producing advanced architectures, algorithms, and software; these innovations enable new products and services (e.g., multimedia applications, blended service infrastructures, and service hosting systems) for the emerging digital TV markets. The project initially created service prototypes that blended IPTV and IMS-based telephony services. These service prototypes were demonstrated in trade shows and generated wide interest, which has led to lab and field trials and has also led to IMS/IPTV product development. The project then focused on blends of cable TV and telephony services. These blends were generally viewed as the first combination of IMS and OCAP technologies. The project will next shift its focus to extend digital TV to include Internet TV. Some of the most intriguing possibilities in this space are services based on the combination of Internet content and delivery with TV display and control.

Network Convergence and Voice over IP (VoIP)

Future networks will have a structure significantly different from today's networks. They aim to converge both wireless and wireline, as well as voice/video and data applications. Services and call control will be separated and spread across many distributed elements, with most features being implemented on dedicated application servers. For successful deployment of such network structures, issues such as reliability, end-to-end delay, manageability, availability, and ease of deployment will have to be addressed. As more and more service-intelligent applications are added to the networks, it will become increasingly important for carriers to orchestrate the interaction of these applications. This project invents, analyzes, and prototypes novel concepts related to networking protocols and infrastructure servers for next-generation communication networks. The focus is on developing software solutions that are used to build a scalable infrastructure for the creation and deployment of converged communication services. Work areas include signaling and networking edge protocols, services control, next-generation application servers, and architectures for integrating Internet and Web technology with traditional voice services. The focus is on protocols such as the Session Initiation Protocol (SIP) and networking architectures such as 3GPP/IMS.  This project is conducted in the Converged Networks and Services Research Center at Bell Labs in Holmdel, NJ, USA. The results of this work have been embodied in several Lucent products. To learn more about the project, please contact me directly.

Next Generation Content Networking

With the exponential growth of the Internet there is a desperate need for the deployment of scalable networking and servicing technologies. A promising approach in overcoming the Internet's notorious bottlenecks and slowdowns is distributing and moving content to the edge of the network where it becomes faster to retrieve. User requests are transparently redirected to and served from local storage devices. We have developed an architecture and a set of building blocks that allow the creation of custom-tailored content storage and delivery networks. An open API enables third party developers to quickly implement advanced services beyond web caching, such as premium multimedia mail, radio broadcasting, etc., on top of our common content delivery platform. This project has been conducted in the Networking Research Laboratory at Bell Labs in Holmdel, NJ, USA, and it has been embodied in Lucent Technologies' content networking solution. To learn more about the project, please contact me directly.

Caching Techniques for Multimedia Streaming

The Streaming Project involves the design, implementation, and evaluation of a new architecture and a set of mechanisms for advanced multimedia streaming on the Internet. It realizes the world's first dynamic caching scheme for streaming media. Basic modules include advanced media transport and automated self-configuration of distributed system components. The Streaming Architecture provides transparently (to the receivers) support for near-zero start-up latency, improved playback quality, improved VCR-like operations, and further reduction in network and server load compared to pure multicast. Our system is designed in compliance to IETF standard protocols (RTSP, RTP). This project has been conducted in the Networking Research Laboratory at Bell Labs in Holmdel, NJ, USA, and it has been embodied in Lucent Technologies' content networking solution. To learn more about the project, please contact me directly.

Local Group Concept (LGC)

My Ph.D. project has been on reliable and scalable multicast communication on the Internet. I designed and evaluated the so-called Local Group Concept (LGC), a generic set of mechanisms to improve scalability of reliable multicast protocols. Based on these ideas, I specified and implemented the Local Group based Multicast Protocol (LGMP) and the Local Group Configuration Protocol (LGCP), which allow efficient data dissemination to a large number of receivers. Both protocols have successfully been used for worldwide experiments on the MBone involving 40 to 50 machines located at 16 different sites in Canada, England, France, Germany, and the USA.

IPv6/RSVP Research Lab

In 1996, I initiated together with our system administrators the establishment of a local Research Lab for the next generation Internet protocol suite. The Research Lab aims for the development of new Internet technologies and provides a local testbed connected to the worldwide 6Bone. Research started with multicast implementations on top of IPv6 and continued with focus on RSVP. Some interesting work on Label Swapping with RSVP has been done in collaboration with the Laboratoire d'Informatique 6 in Paris, France.

Multimedia Transport System (MMT)

The BerKom-II project (1992 - 1996), which was funded by the German Telekom, aimed for the development and the implementation of an ATM-based Multimedia Transport System (MMT), interconnecting Digital, IBM, HP, SGI and Sun workstations. I have been working as project manager at the Institute of Telematics and contributed to the development and implementation of the transport level protocol named XTP-Lite. The implementation platform has been Digital Unix on Alpha workstations. The implementation features comprehensive support for different Quality of Service (QoS) classes, including QoS negotiation and mapping of MMT-specific QoS parameters on ATM QoS parameters.

"If we knew what it was we were doing, it would not be called research, would it?"
 ~Albert Einstein

"Research is what I'm doing when I don't know what I'm doing.”
~Wernher von Braun

“There is nothing like looking, if you want to find something. You certainly usually find something, if you look, but it is not always quite the something you were after.” ~J.R.R. Tolkien