Date: Mon, 24 May 1999
Please find attached my reviews of projects
9905-1, 9905-3, 9905-(5-13).
As is often the case, I find myself giving more
favourable reviews to topic areas I know least about!
9905-7 and 9905-8 are the two in question.
I found the proposals to be of good quality and to
address, what I believe to be, important areas.
However, technically my knowledge in this area
is limited; so you should take this into account during
your discussions.
On a more general not, the Real-Time Profile given in the Call for
Project Proposals is very specific. I do not believe that the
majority of the proposals seriously address this profile.
The exception is 9905-5. However, the author indicates that he
has no idea how to address the problem!
9905-3 perhaps should be considered as a PAMP proposal. I have
read 9905-2 and 9905-4 but I haven't provided
formal reviews.
Let me know if you need more information about the
proposals etc.
Project Number: 9905-1
Project Title: Flexible Reliable Timing Constraints
Scientific merits
This proposal is based on the premise that application
timing constraints are, in some sense, misrepresented
in the design and implementation of real-time systems (in order to
simplify the schedulability analysis and run-time scheduling
algorithms). The result is an overconstrained system.
Whilst I have some sympathy with this view, it is not new.
Even with simply using the "period and deadline model" there
are many settings of periods and deadlines that meet the
end-to-end transaction deadlines.
However, and more importantly, the proposed solution is
rather vague. It is based on the concept of a "feasibility function".
This supposedly describes the temporal relationship
between an activity's input and output. Unfortunately, we are given
no clues as to what form these functions might take and what their
parameters might be. No discussion is given on how the functions
can be decomposed nor of the potential impact on schedulability analysis
and run-time scheduling.
Furthermore, it is not clear what research methods will be used to
attack these problems. Finally, only lip-service
is given to the impact on reliability.
Of course, the above issues are the subject of the research. However,
I would have expected to see a clearer picture of the
direction that the research will take.
Industrial relevance, experience, participation and synergy
This is clearly an important topic for industry.
The supporting letter is in Swedish so
I am unable to comment on their involvement.
Conformity with the real-time profile of the current call
The issue of heterogeneous and composable systems does not
seem to be the main focus of the proposal.
Mobility; both academia/industry and international
There is the very welcome involvement of Ramamritham.
However, it is not clear exactly the role he will play.
It is indicated that he will act as student co-supervisor but
no indication is given on how that supervision will be performed.
I assume that he will be in regular email contact
and will help guide the research. Even if Ramamritham's role is simply
one of project review during his visit, this is still worthwhile.
Interdisciplinarity
Potentially very little.
Exploitation expectations
The supporting letter is in Swedish so
I am unable to comment on their plans for exploitation.
I suspect that the work will not be mature
enough for immediate exploitation.
Complementary activities and financing, critical mass
OK
Project management
No details given of milestones or deliverables (perhaps this
is implicit in the PhD programme?).
No details are given of the liaison with the Industrial partner.
Quality of presentations
In my view this is a vague proposal.
Matching of objectives and resources
Difficult to determine.
Recommendation (approval, rejection or reformulation):
Rejection or Reformulation
Project Number: 9905-7
Project Title: Methods for Integration of Heterogeneous Real-Time
Services into High-Performance Network
Scientific merits
Although this is not my main field of expertise, I think
this is a good proposal in an important area. Understanding how to
integrate hard and soft real-time communications into
a high performance network is an important topic. Too often we
are told that network bandwidth is so high we do not need to
worry about communication delays.
This proposal has the potential to address the problem of
integrating hard and soft real-time communications at its source:
the hardware design. By considering the interaction between the design
and the communication protocols, I would expect some good insights
into how we can solve some of the main future requirements
for real-time systems which are highly distributed and have
large communication demands.
Industrial relevance, experience, participation and synergy
Good
Conformity with the real-time profile of the current call
Only indirectly I would say.
Mobility; both academia/industry and international
Not sure.
Interdisciplinarity
I would say partially.
Exploitation expectations
The industrial letter of support is in Swedish,
so I can't tell. However, I would say there would
be potential if the project is successful.
Complementary activities and financing, critical mass
OK
Project management
OK
Quality of presentations
Good
Matching of objectives and resources
Seems OK.
Recommendation (approval, rejection or reformulation):
Approval. Although I would feel happier if someone with
more expertise on hardware communications reviewed the proposal.
Project Number: 9905-8
Project Title: Real-Time Mobile Communications
Scientific merits
Although this is not my main field of expertise, I think
this is a reasonably good proposal. It is clear that mobile
communications is a crucial topic which potentially will
have an enormous impact on distributed real-time systems.
Hence from a strategic viewpoint this work is very important.
Industrial relevance, experience, participation and synergy
Yes, however I am unable to judge the quality
of the applicants.
Conformity with the real-time profile of the current call
Indirect.
Mobility; both academia/industry and international
OK
Interdisciplinarity
Potentially
Exploitation expectations
I have not seen the letters of support.
Complementary activities and financing, critical mass
OK
Project management
A little weak.
Quality of presentations
A little weak.
Matching of objectives and resources
Unable to judge.
Recommendation (approval, rejection or reformulation):
Given the strategic importance of this work,
I would give approval. However, I am unable to judge the
quality of the applicant. There is clearly some
expertise in parallel computing, it is hard for
me to tell if they have the necessary experience
to make a real contribution. However, I note that the
funding requested is for students already in post
and, therefore, I am more inclined to give my support.
I would feel happier if someone with
more expertise on hardware communications reviewed the proposal.
Project Number: 9905-11
Project Title: Testing of Event-Triggered Real-Time Systems
Scientific merits
This application is based on the hypothesis that
the testability of event-based real-time systems can
be improved by applying constraints on application behaviour.
Whilst in some sense this is obviously true, the three constraints
proposed are rather novel and interesting. However, it is
not clear to me that these constraints bring significant benefits,
or if if they do, it is only by such severe restrictions on
the application that the advantages of event-based systems are lost.
The material referenced to support the claims are a conference
paper and a masters thesis.
Industrial relevance, experience, participation and synergy
Good
Conformity with the real-time profile of the current call
Indirect
Mobility; both academia/industry and international
Not clear.
Interdisciplinarity
Potentially.
Exploitation expectations
No letters available at time of review.
Complementary activities and financing, critical mass
They appear to be mainly a real-time database group.
Project management
OK.
Quality of presentations
OK
Matching of objectives and resources
OK
Recommendation (approval, rejection or reformulation):
I think there is a risk in allocating 3 PhD students
to a project whose theoretical foundation has yet to
be clearly demonstrated. I would have like to have
seen more international review of the work and
a journal paper published.
I would support one student to work on the theoretical
foundation and encourage them to reapply next year for the other
two (when they can demonstrate a firm foundation).
Project Number: 9905-13
Project Title: Functional Integration and Interference
in Embedded Control Systems
Scientific merits
In many industries now, we are seeing a movement away from federated
systems to integrated systems. At first I thought that these prop
knew nothing about major international activities such as Integrated Modular
Avionics (IMA). However, apparently they do as they mention it in passing.
I would have thought that the experience gained from IMA
would have a major impact on this project and would have expected
much more discussion in this area.
Having said that, it is clear that avionics is a different ball game from
the automotive industry, and the solutions that work in
one might not work in the other.
Industrial relevance, experience, participation and synergy
Good, although I would like to see more interaction with a Computer Science Department
to ensure they are not reinventing the wheel.
Conformity with the real-time profile of the current call
I guess.
Mobility; both academia/industry and international
Unclear
Interdisciplinarity
Good, although I would like to see more interaction with a
Computer Science Department.
Exploitation expectations
I haven't seen the supporting letters.
Complementary activities and financing, critical mass
OK.
Project management
It is not clear what each PhD is doing.
Quality of presentations
OK
Matching of objectives and resources
Difficult to determine.
Recommendation (approval, rejection or reformulation):
I would approve 1 PhD student to work on the control theory aspects.
