European Network of Excellence (E-NEXT) Workgroup 1 1st Task Force Meeting on Mobile Ad-hoc Networks (May 19th, 2005, Braunschweig/Germany) ================================================ Participants: - Sandrine Calomme Université de Liège, BE - Tiago Camilo University of Coimbra, PT - Karoly Farkas ETH Zürich, CH - Dan Grigoras University College Cork, IE - Habib-ur-Rehman Technische Universität Braunschweig, DE - Ben Hui Intel Research & University of Cambridge, UK - Thomas King University of Mannheim, DE - Gábor Németh Ericsson, HU - Georgios Parissidis ETH Zürich, CH - Yacine Rebahi FOKUS, DE - Matthias Transier University of Mannheim, DE - Oliver Wellnitz Technische Universität Braunschweig, DE - Lars Wolf Technische Universität Braunschweig, DE - Xiaoyuan Gu Technische Universität Braunschweig, DE Minutes: Oliver Wellnitz (notes taken by Oliver and Xiaoyuan) Agenda: ------- 1. Session 1 (Ben, Yacine) 2. Session 2 (Karoly, Oliver, Dan) 3. Session 3 (Sandrine, Thomas, Georgios, Tiago) 4. Discussion Slides: ------- The slides of this meeting can be found at http://www.ibr.cs.tu-bs.de/events/enext05/ Session 1 --------- Ben Hui: Pocket Switched Networking and its Preliminary Mobility Study ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - Mobile devices communicate and carry data for other people. - Scenarios: delivery service, ad-hoc Google, file sharing - Test-bed: iMotes: small devices with ARM processor, bluetooth and flash mem. - Broadcast searches which are carried by others through the network - Distributed 54 Intel Motes to carry around at a conference - More connections to external devices than to iMotes - Power law distribution for inter-connect time (not exponential decrease as in the random waypoint model) - Large contact time is rare (<=0.1%) - Mobility models needs to be resigned - Stateless forwarding does not work - Future work: Continue mobility measurements, mathematical analysis, create representative mobility models Q(Karoly): Artificial environment, conclusion from a single experiment A(Ben): We actually did three experiments (Students, Office and Conference) Q(Dan): What is the cost of your model? Energy-saving? A(Ben): Just preliminary studies for this project. Yacine Rebahi: SAFE in a Nutshell ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - Ad-hoc networks normally assume non-hostile environment - Routing, resource consumption, and no other nodes have the incentive to forward packets for the others, - Use reputation system (SAFE) to force cooperations and build trust in the networks - Reputation is quantified and bounded to the number of packets forwarded, based on personal perceptions and recommendations from the neighbours - Reputation is propagated only to the neighbours - Malicious nodes are punished (malicious: drop packets excessively) - Protocol: SAFE header, for the reputation info, epidemic methods are used - A history weighting algorithm is used for computation of reputation - Future extensions: new path establishment when a malicious mode occurs, packet monitoring techniques, comparison with CORE or CONFIDANT. Discussion between Gábor, Sandrine & Yacine about selfish vs. malicious nodes and punishment for malicious nodes Session 2 --------- Karoly Farkas: MUGAN - Multi-player Games in mobile Ad-hoc Networks (Part I) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - Motivation: automatic collaboration support, quick service deployment and management, entertainment and education, no central management, heterogeneity, mobility, are the main difficulties. - Target: reusable service provisioning middleware, and game architecture for MANET, game application example, integration with the established infrastructure, context awareness in such scenario, the business model, etc. - Expectations: service provisioning framework, new ways of social interactions, and new income for the developers Oliver Wellnitz: MUGAN - Multi-player Games in mobile Ad-hoc Networks (Part II) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - Pre-game, in-game, and after-game phase - The architecture: generic or application specific? - The approach: distributed zone-servers. - Challenges: choice and number of servers, placement of the servers, mobility management and adaptation algorithms, - For data replication: consistency and efficiency, the strategy of replication, the delivery mechanism, generic replication primitives. - Routing and data transport: game specific problem of routing in MANET game, latency is the concern, generic solution. QoS insurance, L<=150ms - Network partitioning and merging - Integration of fixed-infrastructure networks Dan Grigoras: Service Driven Mobile Ad Hoc Networks Formation and Management ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - The motivation: resource and service sharing - Concerns: zero configurations, IP-address is the constraint of mobility. - net_id: computed by the device that creates the network, it has a NetTTL. - NetTTL: the estimated existence time of the network. NetTTL is refreshed by the network nodes. Discussion about simulation network size and oscillation problems. Session 3 --------- Sandrine Calomme: Overlays for MANET ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - g(n,r(n), looking at asymptotically connected - take an infinite number of graphs. - When in doubt, take (D)ensity x R^2 >=1 for a.a.s. connected Q(Ben): What kind of applications are expected to use overlays? A(Sandrine): Many Q(Georgios): How do you find out about density? A: (Discussion) Georgios Parissidis: Towards a Novel Transport Protocol for Mobile Ad Hoc ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Networks and Multi-Path Routing ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - Single path routing protocols V.S. multipath routing protocols. - The later has more overhead, and is more complex - Multipath routing and TCP: the performance is the concern - Multipath routing being taken into account for the design of the transport protocols - The ad-hoc multipath transport protocol. It cares about the flow and congestion control, reliability Thomas King: Sensor Fusion: A Technique to Improve Location Systems ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - GPS: 4 satellites and line of sight needed - RADAR: wireless LAN APs are needed to determine position mobile device, accuracy: 2.5m - Bluetooth: 10 meters - Combine data from several sensors to improve accuracy (Sensor Fusion) - Generic Location System Architecture (Bayesian Network) Tiago Camilo: Wireless Sensor Networks in 4G Integration ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - No standard architecture, low bandwidth, low energy consumption - MAR - Mobile Advanced Routing - IPv6 Anycast address for different services (Sound, Humidity,...) - SICS and the IETF work Discussion: ----------- - Ericsson: We need a business model to sell a project to management - We are interested Traffic models - There are a lot of applications with sensor networks - Ad-hoc network infrastructure - Set up an AP in the train to help data transmission - Discussion about feasibility, business models - What applications do we expect in ad hoc networks? - Multiplayer Games, Chat, Dating Service for Commuters, Filesharing - Possible cooperation / Similar research interests: Tiago <-> Dan: IPv6 concept Oliver: ETHZ, Ericsson, Liège Ben <-> Dan: Reputation System - Discussion about Application, Mobility Models, Scenarios - Use the mailinglist more often - Write whitepaper