Home
About
Steering Committee
Research Network
Events
Human Mobility
Papers and Reports
Useful Links
Administrative
Mailing Lists
Announcements
Contact
Members Only
SC Only

 

Minema Summer School

MiNEMA summer school, July 11-15, Klagenfurt University, Austria

Abstracts from the talks
The Publish/Subscribe Communication Paradigm and its Application to Mobile Systems
Roberto Baldoni, Universita di Roma 'La Sapienza', Italy

Pub/subscribe communication is an appealing many-to-many abstraction to disseminate events in a distributed and mobile system. Implementations of this abstraction have a main target: avoid the flooding of a network connecting publishers and subscribers while keeping very high the level of reliability of delivery of events to subscribers. The class studies the basic architectures of publis/subscribe communication systems. These architectures will point out basic mechanisms that can be used to achieve that target. These mechanisms allow to classify different systems and prototypes according to the way they propagate and match events and subscriptions. The class then examines a few well-know pub/sub systems aiming to frame such a systems into previous classification. Finally the impact of mobility is analyzed in terms of instability of the underlying network, power consumption and reliablility of event delivery.

slides (ppt) (pdf)
Open Middleware Architecture and Adaptation
Gordon Blair, Lancaster University, UK

As a mobile devices moves it naturally encounters change in its environment, i.e. change in context (such as the device location), change in network conditions and change in available resources. Therefore, mobile middleware must be able to adapt its behaviour dynamically to continue to provide the best level of service to applications. In this chapter we focus on the use of reflection to build dynamic middleware. A series of case studies illustrate solutions that have successfully benefited from the reflective middleware approach.

Reflection chapter (pdf)
Multimedia Streaming in Heterogeneous Networks
Laszlo Böszörmenyi, Klagenfurt University, Austria

We discuss the inherent contradiction between the requirements of multimedia streaming and the actual capabilities of networks and operating systems. Actual resource management in networks and operating systems is optimized for the best-effort strategy; video streaming requires, however, quality-of-service aware strategies.

slides (pdf)
Group Communication for Mobile Environments
Roy Friedman, Technion, Israel

Group communication middleware provide services that facilitate the development of distributed applications consisting of multiple participants that need to exchange information among themselves. In particular, group communication offer services such as consistent membership notification, automatic failure/recovery detection, reliable multicast, ordered delivery, flow control, etc. Group communication systems have been developed initially in the context of LAN oriented clustering. Indeed, group communication based clustering products play a central role in an increasing number of mission critical and business critical applications, including stock exchanges, air-traffic control, data centers of large retail chains, nuclear plant, energy companies, financial institutes, highly available Web server farms, as well as highly available and scalable clustered file systems. As many abstractions, group communication brings with it a tradeoff between semantics and performance. In particular, when applying group communication to mobile computing environments, these tradeoffs become more acute, and often group communication middleware developed for wired LAN environment does not function well in a mobile setting. At the same time, as been proven by projects performed by students in our lab, group communication middleware provide a very powerful programming paradigm for mobile applications as well. Examples for such applications include, e.g., ad-hoc auctioning systems, distributed interactive games, distributed collaborative tools, serverless video conferencing, collaborative caching, and automatic offloading of computation from weak devices to stronger ones. All these applications become simple to implement with an appropriate middleware! In this talk I will give an overview of group communication, what they are, what they offer, what are these performance vs. semantics tradeoffs, and analyze what can and should be done in mobile systems. In particular, I will present the unique challenges of mobile environments and ways of addressing them. This will include, as time permits, both mobile ad-hoc networks and cellular networks. Finally, I will present some open problems for the middleware research community.

slides (pdf)
Middleware for Mobile Computing
Mads Haahr, Trinity College, Ireland

This talk will address the motivation for middleware support for mobile computing and the basic issues to be addressed by such middleware. It will consider possibilities for adapting traditional middleware paradigms for use in mobile settings versus the need for new middleware paradigms. It will examine techniques for adapting existing middleware architectures, especially those based on method invocation, for mobility and briefly consider other more asynchronous paradigms.

Mobility and its Effects on Network Performance
Zygmunt Haas, Cornell University, USA

We usually think about mobility in wireless networks as an impairment; a penalty that the network provider and its users have to 'pay for' to support tether-less access. After all, mobility is responsible for fading of radio signal, which, in ad hoc networks, eventually leads to path breakage, and to the need to reroute a connection. In cellular networks, mobility requires call handover between base-stations and possibly between mobile switching centers. Mobility requires the system to re-authenticate a user and allows user to mount security attacks more easily. But have you ever thought that mobility can actually have beneficial results on the operation and the performance of wireless networks. Could you imagine that mobility can actually increase the network capacity, improve the network reliability and its availability, and in some cases make the network actually more secure? Well, it can - sometimes. In this presentation, I will attempt to expose both, the adverse and the favorable effects of mobility, and discuss ways to cope with the adverse and to exploit the favorable. In particular, I will discuss how mobility is modeled in the scientific literature and what the shortcomings of these models are, shortcomings that can transform results of studies into meaningless statements. I will discuss how replacing one mobility model by another can easily turn conclusions upside down. Finally, I will ponder on whether there is any hope that a realistic mobility model, one that is feasible to use with the current state-of-the-art simulation tools, will be available to researchers and network designers.

Secure and Cooperative Wireless Networks
Jean-Pierre Hubaux, EPFL Lausanne, Switzerland

As wireless networks are becoming more and more pervasive, users will increasingly rely on them. At the same time, the devices are becoming more and more programmable, thereby paving the way to malicious and selfish behavior. In the first part of this tutorial, we will review some basic techniques of security and cryptography; we will then show how they can be adapted to prevent malicious behavior in wireless networks. We will focus in particular on key establishment techniques, and consider the case of vehicular networks. In the second part, we will briefly introduce game theory. We will then show how this formalism can be used to study selfish behavior in wireless networks. In particular, we will consider the problem of packet forwarding in self-organized multi-hop wireless networks.

slides (pdf) part 1 part 2 part 3 part 4 part 5
Introduction to Ad Hoc Network Routing
David B. Johnson, Rice University, USA

In an ad hoc network, mobile nodes such as notebook or handheld computers cooperate to forward packets for each other, without the aid of network infrastructure such as access points or base stations. To allow nodes not within direct wireless transmission range of each other to communicate, a route between them through one or more other nodes in the ad hoc network must be known. As nodes in the network move or as other factors such as node failures or wireless propagation changes occur, existing routes may break and new routes must be learned. This presentation will provide an introduction to the problems and solutions in routing protocols for mobile ad hoc networks. For unicast packet routing, ad hoc network routing protocols are often classified as either proactive protocols--in which each node always attempts to maintain current routes to all possible destinations--and reactive protocols--in which a node dynamically attempts to discover a route to some node only when it has a packet to send to that destination. We will look at examples of both types of protocols, the tradeoffs between the two types, and ways in which they can be combined to form hybrid routing protocols. We will also look briefly at protocols for routing multicast packets in ad hoc networks.

Ad Hoc Networking in the IETF
Charles E. Perkins, Nokia Corporation, USA

Within the IETF, the working group for Mobile Ad Hoc Networking [manet] has recently made new steps towards standardizing new routing protocols. In particular, there is now a document specifying a new protocol called DYMO (for Dynamic Mobile Networks). I will discuss these recent steps forward, and to give concrete examples I will describe more specifically some recent improvements to 'Ad Hoc On-Demand Distance Vector' protocol (AODV) [RFC 3561]. Four protocols have been published as experimental specifications within the [manet] working group of the IETF. I will describe in brief certain aspects of these protocols, and then go into a more detailed description of DYMO. After describing these protocols, I will then give some opinions about how a convergence may be effected. Convergence between protocols does provide a significant force for acceptance, and so technology that finds commonality between otherwise divergent protocols is highly desirable. The good and recent news is that commonality has been identified, so that link-state and distance-vector protocols can be merged, as well as (possibly!) proactive and reactive protocols. Along the way, much has been learned about the basic nature of routing protocols, and yet much remains to be learned.

slides (pdf)
Recent Developments in Middleware Standardization for Mobile Computing
Kimmo Raatikainen, University of Helsinki, Finland

Abstract to be posted

slides (pdf)
Gossiping in the mobile world
Luis Rodrigues, University of Lisbon, Portugal

Gossip-based algorithms, also called ``epidemic'' or ``probabilistic'' broadcast algorithms, are protocols where nodes randomly contact other nodes to promote the fast dissemination of data. Gossip protocols are extremely resilient to failures of participants and to changes in the underlying network topology, which makes them quite appealing for highly dynamic settings such as mobile and ad hoc networks. Gossip-based algorithms have been applied to a variety of different uses, from monitoring, management, and data mining to the support of large number of participants in multi-user games. This talk will introduce gossip-based algorithms and identify their strengths and weaknesses, namely when compared with flooding or tree-based approaches. Subsequently, the talk will concentrate on the use of gossip-based algorithms in mobile and ad hoc networks for different tasks, such as data dissemination and routing. The most significant approaches will be illustrated using concrete algorithms. We will conclude by pointing open issues and future research directions.

slides (pdf)