OPENSIG '99 Workshop Open Signaling for ATM, Internet and Mobile Networks October, 14-15, 1999, Carnegie Mellon University, Pittsburgh, USA

Talk Abstracts

Programming the Physical Layer in Wireless Networks

Many of the limitation of today's wireless communications systems, such as multiple incompatible standards and the inability to dynamically optimize to the mobile environment, result from a lack of flexibility in the physical layer. Software radio, the implementation of wireless physical layer functionality in software, provides the flexibility needed to overcome these limitations.

This talk describes a software radio architecture which is based on wideband digitization, a general purpose processor and application level software. Using this architecture any aspect of the physical layer, including channel selection, coding and modulation to be dynamically re-programmed. This enables a wireless network to dynamically adapt to changes on the environment, traffic or user requirements, and compile the best radio for the current conditions.

Programmable Handoff

We argue that future wireless access networks should be built on a foundation of open programmable networking allowing for the dynamic deployment and management of new mobile and wireless services. Customizing handoff control and mobility management in this manner calls for advances in software and networking technologies in order to respond to specific radio, mobility, and quality of service requirements of future wireless service providers. Two novel applications of programmable handoff are presented: (i)
the capability to support multiple styles of handoff over the same physical infrastructure; and (ii) the capability to allow mobile devices to roam between access networks with different signaling systems for mobility management

Active and Ad-hoc: Experiences with Audio Streaming in a Wireless Active Network

We have implemented an active ad-hoc routing protocol similar to cellular IP in a two-layered active network architecture. Active packets proactively monitor a set of wireless nodes, these packets also setup and maintain a default routing infrastructure and create data delivery paths which are attached to the default routing core. This defines a `network personality' in which passive data packets (containing delay sensitive audio data) are shipped. Forwarding takes place in the lower layer built around the Simple Active Packet Format (SAPF) and is implemented inside the Linux kernel, while active packets are processed in user space.

In this talk we explain our active routing protocol, the experiences we gained regarding its dynamics, and some companion active diagnostic applications like the online topology tracking and visualization.  This work was done in the joint Uppsala University and Ericsson Switchlab research project calledARRCANE (Active Routing and Resource Control in Ad-hoc NEtworks).

Network Resource Information Services for Adaptive Applications

Network-aware applications adapt their resource demands in response to fluctuations in the availability of resources.  Such applications therefore need to find out what resources are available, must be able to identify changes, and if possible are able to estimate future resource availability. Each of these aspects, however, is far from easy to manage since the overhead to obtain resource information must be small -- otherwise an application may not realize any benefits from adaptivity. Furthermore, an application's agility imposes another constraint:  if an application can adapt every few seconds, it may be futile to obtain resource information at a much higher frequency.  The Remos (Resource Monitoring System) provides basic resource information services, and we use this system to illustrate various tradeoffs.

A System for Flexible Network Performance Measurement

NIMI (National Internet Measurement Infrastructure) is a software system for building network measurement
infrastructures.  A NIMI infrastructure consists of a set of dedicated measurement servers (termed NIMI probes) running on a number of hosts in a network, and measurement configuration and control software, which runs on separate hosts.  A key NIMI design goal is scalability to potentially thousands of NIMI probes within a single infrastructre; as the number of probes increases, the number of available measurable paths increases via the N-squared effect, potentially allowing for a global view of the network.

A fundamental aspect of the NIMI architecture is that each NIMI probe reports to a configuration point of contact (CPOC) designated by the owner of the probe system.  There is no requirement that different probes report to the same CPOC, and, indeed, there will generally be one CPOC per adminstrative domain participating in the infrastructure.  But the NIMI architecture also allows for easy *delegation* of *part* of a probe's measurement services, offering, when necessary, tight control over exactly what services are delegated.

The architecture was designed with security as a central concern: all access is via public key credentials.  Each NIMI probe is configured by its CPOC (or a delegatee of the CPOC) to allow particular sets of operations to different credentials.  The owner of the probe can thus determine who has what type of access by controlling to whom they give particular credentials.

The sole function of a NIMI probe is to queue requests for measurement at some point in the future, execute the measurement when its scheduled time arrives, store the results for retrieval by remote measurement clients, and delete the results when told to do so.  An important point for gaining measurement flexibility is that NIMI does *not* presume a particular set of measurement tools.  Instead, the NIMI probes have the notion of a measurement "module", which can reflect a number of different measurement tools.  Currently, these measurements include traceroute, TReno, mtrace, and zing (a generalized "ping" measurement), but it is simple to include other active measurement tools on selected probes.

In addition to giving an overview of the architecture, we will discuss experiences with running NIMI to conduct a number of Internet measurement studies.

Extensible Architecture for Passive and Active Protocol Interposition

The rate of growth for the Internet has placed a severe tax on the network infrastructure, leaving many resources such as routers and highly trafficed Web servers in a state of constant overload.  Understanding the interaction among the many Internet protocols is a key challenge necessary for its rational growth. Compounding the problem, is that most of the software executing the protocols is "shrink-wrapped" and is not amenable to scrutiny or modification for performance measurement -- a backbone router collapses within seconds with full debugging turned on.  It is precisely at these points where the performance effects of protocol interaction are the greatest, and most poorly understood.

The talk focuses on the architecture of two related tools: Windmill and Protocol Scrubber.  Windmill is an
extensible probe that utilizes passive measurement techniques for eavesdropping on target network protocols.  Windmill enables experimenters to measure a broad range of protocol performance metrics by both reconstructing application-level network protocols and exposing the underlying protocol layers' events.  Protocol Scrubber is active an programmable mechanism for explicit on-line monitoring and enforcement of network security policies.  The scrubber is a transparent interposition mechanism for the conversion of ambiguous network flows into well-behaved flows both at transport and application protocol layers.  Key concepts underlying the architecture of these tools: protocol interposition; abstraction-breaching event monitoring; protocol reconstruction; and extensible experiment engines.  In addition to presenting the architecture of Windmill and Protocol Scrubber, the talk will highlight our experiences using these tools in several experimental studies.

Using Internet Measurements to Direct Clients to Servers

This talk discusses the problem of choosing the right web server to direct a client to when servers on many backbones are available.  The talk will examine the network conditions that make one server better (or worse) than another, and methods for detecting those conditions.

Capsue-based Active Networks: What Have We Learned?

At least eight active network prototypes have been built over the past three years. What have we learned from them? In this talk, we reconsider the active network vision in light of our experience with one of these prototypes, ANTS. We do this by comparing what we have learned with the original vision in three areas
that characterize a "pure" active network: the capsule model of programmability; the availability of that model to all users; and the motivating applications. We argue that we have made substantial progress towards providing a more flexible network layer while at the same time addressing the performance and
security concerns raised by the presence of mobile code in the network. At the same time, we have somewhat modified our positions compared to the original vision.  This talk will discuss our findings and their implications for ongoing research.

Scalable Fair Reliable Multicast Using Active Services

Experiences with Active Kernel Modules

In this talk we will present a classification of possible Active Applications.  Using this classification we will
provide some justification for active kernel modules as one approach to active networking. As an example, a video application using an active kernel module to perform congestion control will be presented.  We will
conclude by relating our experiences to date with active networking in the kernel domain.

New Models and Algorithms for Active Networks

In todays IP networks most of the network control and management tasks are performed at the
end points. As a result, many important network functions cannot be optimized due to lack of sufficient support from the network. The growing need for quality guaranteed services brought on suggestions to add more computational power to the network elements.

This paper studies the algorithmic power of networks whose routers are capable of performing complex
tasks.  It presents a new model that captures the hop-by-hop datagram forwarding mechanism deployed in todays IP networks, as well as the ability to perform complex computations in network elements as proposed in the active networks paradigm.  Using our framework, we present and analyze distributed algorithms for basic problems that arise in the control and management of IP networks.  These problems include:  route discovery, message dissemination, topology discovery, and bottleneck detection.

Our results prove that, although adding computation power to the routers increases the message delay, it shortens the completion time for many tasks.  The suggested model can be used to evaluate the contribution of added features to a router, and allows the formal comparison of different proposed architectures.

The Darwin Router Control Interface

The CMU Darwin project has developed a set of customizable resource management mechanisms that allow networks users (e.g. service providers) to directly control the resources that are allocated to them.  One such mechanism is the control delegate, a code segment that is provided by the user to implement customized control functions and protocols on the router.   In this talk we will focus on the Router
Control Interface, the interface that delegates use to control router behavior.  We will describe the functions delegates can perform and the mechanisms that are used to restrict the scope of delegate actions.  We will conclude with some examples of the use of delegates.