A number of Working Groups is established that aim at co-ordinating work in areas that are common among two or more projects. The WGs will ensure that complementarity between projects is maximised and information flows efficiently. The WGs formally meet twice a year, but it is expected that contacts among WG members much more frequent by phone or e-mail. Each WG has a convenor who acts as the contact point, is in charge of the agenda of the meetings, and communicates the conclusions to the WG participants as well as to the SC, via its project co-ordinator.
There are 5 WGs on the following subjects:
J. Van Campenhout
Uni. Gent (OIIC)
In the context of the architectural and demonstrator related aspects of the MEL-ARI Opto projects, the following objectives were put forth:
To collect and exchange more concrete and reliable factual data relating to some relevant geometrical and behavioural properties of state of the art optical interconnects, including:
- area required to realize an interconnection (including driver or receiver amplifiers)
- other geometric restrictions due to the envisaged hybridization process
(e.g. scattered vs. grouped sources/receivers)
- power dissipated (peak, average) on the sender side and on the receiver side
- bandwidth of the interconnection; power-bandwidth relationship for a given BER
- total latency of an interconnection, including driver and receiver delays
- trade-offs between multiplexed and non multiplexed optical interconnects
As all MEL-ARI projects intend to do some development at the technological level, and as the proposed developments are rather diverse (VCSELs, LEDs, different materials, different wavelengths, modulators vs. sources, etc.), the proposed collection and uniform presentation of these data is possible, and should represent a definite added value. Obviously, for this as well as the other goals, data gathering should not be limited to data produced by the actual consortium, but will have to include data published in the open literature or presented at the relevant conferences. There is a clear link with the road map activity.
To gradually identify those architecturally relevant aspects of optical chip-to-chip interconnects (both VCSEL-based and LED based) that provide a quantifiable advantage in comparison with electrical interconnects. In the various project descriptions, several advantages of optical interconnects are claimed or assumed, some of which are of a quantitative nature, and some of which are of a more qualitative nature. Potential advantages of massively parallel optical interconnects, cited in the project descriptions, include the following:
- The ease of realizing a broadcast operation (optical fan-out).
- Nearly skew-free clock distribution (a use of the former).
- Globally lower power dissipation when compared to electrical chip-to-chip interconnects
(in terms of global speed-power product).
- The possibility to realize area I/O (a way out of pin limitations).
- The possibility to end up with shorter average electrical wire lengths (speed advantage? area advantage?).
- Reconfigurability (less quantifiable).
- High aggregate bandwidth with relatively low latency when compared to sequential electrical interconnects.
While most people hold a certain degree of belief in these claims (it is the driving force behind the MEL-ARI projects), solid substantiation in a systems context is badly needed. The developments and assessments proposed in the various MEL-ARI projects will unveil some of the hype associated with optical interconnects, and provide us with a much firmer understanding of the true benefits
Identify, and share views on other, both specific (niche type) and more generic applications where optical interconnects could be beneficial. The results obtained and data collected during the projects in the end should leave us with much clearer quantitative view on realistic applications. Domains include special and general purpose high-performance computing, telecommunications, generic high performance circuitry, etc.
Clearly, reaching these goals requires a very open exchange of ideas and results among the various projects (including the projects without an architectural activity MONOLITH and SIBLE). MEL-ARI projects have pledged to provide a sufficient level of openness to bring the WG objectives within reach. Furthermore, since membership is voluntary, it is up to the individual projects to decide whether, and to what extent, they will participate in the information exchange.
Activities and Results (status per 30.1.1999):
The Architectures Working Group has been set up, and meetings were held as planned in the context of the following MEL-ARI meetings:
During the first meetings, the objectives listed above were presented to the Working Group members, and agreed upon. Furthermore, the different participating projects presented their views and plans with regards to the architectural issures and OE demonstrators. See the overview table for a brief summary of the initial individual project's goals, demonstrator plans, and architectural focus.
During the subsequent meetings, progress made by the projects was reported to the Working Group Members.
In those presentations, two main architectural fields of application of optical interconnect were discussed: The application of optical interconnect as an enhancement of processor-to-memory, processor-to-peripheral, or processor-to-processor interconnects The RODCI project aims at exploiting the specific properties of optical interconnect, in particular the virtually unlimited potential connectivity of optical interconnects, for the realization of dynamically reconfigurable multiprocessor systems.
The basic concept is the dual of traditional dynamic load balancing: when the interconnect network detects hot spots in the inter-processor communication network, the interconnect network topology is changed by means of a local transformation. in which only three nodes are involved at a time. Hence, in contrast to traditional load sharing approaches, there is no process migration among nodes, but rather a migration of processing nodes themselves to the locations where they communicate more efficiently. To allow such dynamic reconfiguration, every node in the system should potentially be connectable to any other node in the system. In large systems, this is deemed feasible only with optical interconnects.
The concept was first introduced, and during the later meetings, simulation results were presented indicating the feasibility and the potential processing speed gains realizable using the dynamic reconfiguration technique.
The application of massive optical interconnects, for the construction of logic level 3D opto-electronic systems
BACK TO WORKING GROUP MENU
Convenor : Karlheinz Gulden , CSEM Zurich (SPOEC)
The VCSEL/MCLED and Drivers Working Group was initiated during September 1996 as part of the MEL-ARI cluster activities. It provides a forum for the exchange of information on recent progress in the design and fabrication of active microcavity devices and driver circuits for optical interconnects between the different cluster projects.
The VCSEL/MCLED and Drivers Working Group is intended to help the MEL-ARI partners working on related problems within different MEL-ARI projects to arrive at optimum design choices for the realization of active light emitter and driver components and complete subsystems in optical interconnects.
In especial, the information exchange on recent technological developments, avaible foundry services, and expected developments will support the integration of arrays of VCSEL/MCLED devices with CMOS driver circuits in future optical interconnect systems. The working group convenor supports the information flow between these groups and organizes the agenda of the WG meetings. Besides technical presentations, strategic issues such as the MEL-ARI roadmap are discussed during the working group meeting .
Within the MEL-ARI Opto Cluster, several groups are working on VCSELs, three groups develop microcavity LEDs and one group is designing driver circuits. Below are some links to internet homepages of working group partners:
|University of Ulm||http://www-opto.e-technik.uni-ulm.de/forschung/index.html|
|University of Sheffield||ftp://ftp.shef.ac.uk/pub/uni/projects/smd/|
CSEM Zürich offers custom VCSEL arrays commercially. For more information please write to (E-mail removed)
|Information on MCLED development can be found under:|
|IMEC Leuven and Gent||click here to download a pdf file|
|Information on the group working on driver designs can be found under|
The working group meets at regular intervals during the MEL-ARI review and workshop sessions.
Next planned activities of the VCSEL/MCLED/Driver WG:
For further informations please contact K. Gulden ( (E-mail removed) )
Convenor: G.Crean , NMRC
Hybridisation is a key technology activity of the OPTO-electronic interconnect cluster of MEL-ARI projects. By its very nature, it is multi-disciplinary in scope, and requires the development of a synergy between the developers of transmission and receiver devices, and the optical interconnect developers.
Establish the technological challenges in the area of hybridisation common to the OPTO cluster of projects.
Compare the different approaches used by each project to address these common problems.
Identify a "best practice" technological approach to focus all projects in the cluster on the best solutions to common hybridisation needs.
Enable the interaction of hybridisation ideas within the cluster.
Respond to requests for assistance in hybridisation technologies, pooling the resources and expertise within the OPTO cluster.
Actions: The following actions and activities are proposed: Each project to name one contact person for hybridisation activities in that project, plus additional interested participants, to the WG3 convenor. Definition of hybridisation problems as seen by each project, to take place at the April 9/10 1997 WG3 meeting within the MEL-ARI Opto cluster meeting. Circulation and discussion of "best practice" proposed solutions by May 1997. Requests for support to be made by projects before end August 1997. Synthesis documents to be produced by end October 1997 and end October 1998. The convenor will propose meetings as necessary and will facilitate communications within the working group, principally by e-mail.
BACK TO WORKING GROUP MENU
Convenor: Hans Melchior ETH Zurich (RODCI & OIIC)
Collect and exchange information and create synergies concerning all aspects of photodetectors relevant to the advancement of "Optoelectronic Interconnects for Integrated Circuits" projects of MEL-ARI OPTO.
Help the MEL-ARI partner projects to arrive at optimal design and realization choices for photodetectors and photodetector arrays suitable for optical interconnects at the intra- and inter-VLSI chip level.
Evaluate different photodetector solutions, drawing conclusions and generating inputs to the road-map activities of MEL-ARI OPTO.
The following actions and activities are proposed:
Invite each MEL-ARI project to comment on present draft of objectives and actions.
Organize interactions concerning photodetector issues amongst the different MEL-ARI OPTO projects: Set-up e-mail list of participants from each MEL-ARI OPTO project in WG4:
Would each project leader please send the names that should be put in this list to convenor of WG 4 by the end of February 1997.
Organization of WG4 meetings:
-Would each MEL-ARI OPTO project please consider contributing to the WG4 meeting on photodetectors and name
- title of their contribution
to the convenor by the end of February 1997.
Produce synthesis documents indicating progress of current work and final synthesis and conclusions of MEL-ARI OPTO projects concerning Photodetectors in inter and intra chip optical interconnects.
BACK TO WORKING GROUP MENU
Supelec, Metz (SPOEC)
Projects participating in the WG :
A. Neyer, Univ. Dortmund and H. Thienpont, VUB
S. Paineau, Thomson CSF LCR
M. Goetz, Supelec and A.C.Walker, HWU
This is a rather small WG and it appears that the three projects involved in it are very different and consequently there is limited overlap.
In the OIIC project, integrated circuits are to be interconnected by means of PMMA Optical Pathway Blocks. These components can either make use of free space or of guided wave flexible guides.
In the RODCI project, the light is also guided from the source to the detectors and diffracted by Computer Generated Holograms (CGHs).
In the SPOEC project, free space propagation is used, but also CGH techniques to fan out input beams to the modulators and in order to replicate the array of input signals.
Despite these differences, it seemed to us that it would still be worth going on with this WG, so that the advantages and disadvantages of the various techniques being used can be compared. The first meeting will be held next Spring, during the First Review Meeting in Brussels.
At this early stage, there has been very little information exchange between the projects. However, as said during the last Steering Committee, the WG's are the ideal place for technical interaction to occur across the projects.
BACK TO WORKING GROUP MENU
The URL of this page is /esprit/src/melop-wg.htm
It was last updated on 21 May 1999, and is maintained by P. Malinverni - (E-mail removed)