Minutes of ALPINE Kickoff Meeting IC DOC 28 June 1999. -1. Agenda was agreed 0. Introductions M Sloman mss@doc.ic.ac.uk D Hutchson dh@comp.lancs.ac.uk L Mathy laurent@comp.lancs.ac.uk R Cardoe cardoe@comp.lancs.ac.uk D Waddington dan@comp.lancs.ac.uk Jim Hardwicke jim.hardwicke@bt.com Hamid Gharib hamid.gharib@bt.com N Damianou ncd@doc.ic.ac.uk Jeff Magee jnm@doc.ic.ac.uk Paul Mckee paul.mckee@bt.com Mike Fisher mike.fisher@bt.com Glen Maclarty marty@socs.uts.edu.au Atanu Ghosh atanu@socs.uts.edu.au Michael Fry mike@socs.uts.edu.au Naranker Dulay nd@doc.ic.ac.uk Dan Chalmers dc@doc.ic.ac.uk Nat Pryce np2@doc.ic.ac.uk Ian Marshall marshall@drake.bt.co.uk Jon Crowcroft jon@cs.ucl.ac.uk apols from Luis Velasco velascol@drake.bt.co.uk ACTION: Jon to set up and maintain web page and maillist. Jon gave an introductory talk (see slides on project web page at http://www.cs.ucl.ac.uk/research/alpine for details Morris noted that several people present would be at IWAN for the next couple of days; also that the EPSRC managed program on prograsmable networks for telecoms may not happen unless there is more industry and Software Engineerign sign up. lastly that UCL, Lancs and IC and BT are also bidding on an IRC in the area of mobility in medical care, and that there could be good synergy between that and this project. 1 Process: Ian Marshall then outlined the way he would like the project to proceed. Work BT wil be a full partner, indeed have more effort than the academic institutes in this project. Process a) Partners should arrange site visits 2-3 times per year for a week or so to BT _ perhaps these could be organised in pairs (esp. to coinicide with UTS visits) to get 3 way interactions b) output in the form of papers was to be encouraged - in fact the contract says "2 papers pe year, and 1 s/w drop per year". ACTION ALL: check if the cotnract has arrived! Report back to Ian Marshall on status c) We should meet in "plenary" about 1 per year, but perhaps we could also have an annual workshop (more on this later). Could also make use of phone teleconferences between partners as needed d) BT will do some direction settign as neded e) There are some basic project management services needed i) Web (done) http://www.cs.ucl.ac.uk/research/alpine ii) maillist (done) alpine@cs.ucl.ac.uk iii) CVS store - tbd, unilaterally at first iv) workshops - see later 2. Technical presentations There were presentations by UTS - Mike Fry outlined the state of the art on the DPS work, and talked about areas he'd like to move to such as mobility - he gaev a nice scenario of a palm user on a plane traveling over various regions, using DPS ALAN services on the plane, and off in orbit, on the ground, at sea and so on BT - Ian outlined the DPS extension work they'd done including policy extensions using LDAP and basic routing (could use URNs), and source and constraint based route selection. IC DOC - Morris and Nat outlined work on newer policy languages, and types of policies (e.g. delegation and filtering, as well as composite (e.g. role) policies). Nat outlined the previous work on composable interactions and protocol stacks, and talked about future extension to more dynamic systems - some of this will be in conjunction wth nomadic and dynamically confgured systems EPSRC project which Dan Chalmers and Southampton are working on with HP. UCL CS- Jon talked about possible name, address and routing techniqies for ALAN - firstly, using LDAP as a mapping service, and URNs as the "address" of an ALAN DPS node, we can envisage a multi-path, multi-metric, dynamic distributed routign protocol (two actually) 1/ for unicast:- based on Dual with elements of ifdeas from NIMROD and PNNI or other diffusing computation, but using path vector isntead of destination (advantage - path can be source selected, and can include sources) 2/ for multicast, uising destiantion sequences ad hoc distance vector multicast, extended for paths again, to get rapid converging ad hoc trees..... Morris suggested we probably need hieratrchy to get more scaling - Jon agred, bu t said we sytill need transparency since a source needs to specify full path - can use source demand routes with caching for this though... Lancast - David Hutchison talkjed about 3 areas of Lancaster work i) building group managemenbt and qos (e.g. of application - colalborative browsing) ii) Performance comparison between ALAN and active node approach & Architecture comparison iii) Support for media processing (this is background and will probably happen in the other project they have on MPEG 7 versus XML) 3. Details Infrastructure - we agreed we can use ordinary Internet connectivity, as well as standard JACA releaseds for now. Contract - without seeing, it, we had a brief debate on IPR and problems that might arise if the Unversities saw the wrong overheads for BT having 100% ownership. Workshop - DH is going to check out the possibility of us running OpenSIG (hosted at UCL maybe) in 2000 - september timeframe - would probably be joint with IEEE PIN. IWAN - Ian Marshall is going to check out whether we mihgt bid for IWAN 2001 Our own workshop (or maybe EPSRC involvement) could be reserved for 2002 near end of project? Detailed workplan - BT insisted we need to deliver a detailed plan (at least a page of A4 ) to them by August 7th - Jon suggested it have a placeholder for pre-requistites on it. It should list papers, visits, technical work and s./w outputs. There was a brief discussion about use of CVS - We are not integrating anything yet, but ACTION: UTS will ship (publically) the URL for the older version of DPS, and privately, the new version of DPS asap, to get people up and running. AOB None Time and Date of Next Meeting Sep 24, 1999, 1 hour from URI meeting, UCL CS April 3, 2000, at BT Labs