You are currently browsing the category archive for the ‘opinion’ category.

DARPA has an interesting program called META that looks at approaches for component based design of complex engineering systems, especially aerospace systems. The idea is interesting and there is definitely already a large and growing number of papers being published at various real-time systems and control conferences on this systems engineering aspect. Yet I must say that I haven’t found a good paper on this subject for systems that involve any kind of analog component. Aviation Week has an article on the topic today that mentions the analogy with hierarchical and component-based automated design of integrated circuits. However, it is well known that these approaches work well with digital circuits but that as soon as analog signals enter the picture, automated design becomes much more complicated. I do not think we currently have proper abstractions necessary for component-based analog or mixed-signal system design. Moreover, these abstractions are likely to be more complex to develop than thinking of a transistor as a switch for digital design. For example, analog component-based design will have to trade-off performance for robustness of the components and convince the system designer that the overall system performance is still acceptable in interesting cases.

Update 11/02/10: Aviation week has another interesting article on systems engineering and integration, which mentions the META program.


Automatic Dependent Surveillance-Broadcast (ADS-B), a satellite based tracking system for aircraft (much like a GPS), is supposed to replace radars in NextGen, the Next Generation Air Transportation System (the current plan is to maintain radars as a backup I believe). Yesterday, the FAA issued performance requirements for ADS-B, which most planes must have by… 2020! It’s a fact that pretty much everything related to the overhaul of the national airspace system seems painfully slow [1].

ADS-B is an important technological component of NextGen, as it will allow more people, including the pilots, to have better information about the current state of the system. In fact, commentaries often equate ADS-B with NextGen, although according to the FAA, NextGen is just an “umbrella term for the ongoing, wide-ranging transformation of the National Airspace System (NAS)”. The ultimate goal of NextGen is of course not  simply track aircraft better. Rather, it should improve safety, reduce air traffic congestion and hence result is lower fuel consumed and shorter flight times, etc. To control the National Airspace System, it is clearly beneficial to have better sensors such as the ADS-B system. Ultimately however, this will not be sufficient to improve the performance of the system: better large-scale decision and control algorithms that use this information will need to be accepted and applied by the air traffic controllers and airlines.

[1] J.R. Wilson, “NextGen: A Slow Transformation”, Aerospace America Magazine, May 2010.

I’m coming back from CPS Week, which I attended for the first time. The nice part was that the conference took place at KTH in Stockholm. The not so nice part was that I got stuck (like everybody else there) on my way back due to the volcanic eruption in Iceland. Fortunately I only wanted to reach Paris, not the US. That took me 3 days rather than 3 hours however.

The conference was fun. For those who wonder, CPS stands for “Cyber-Physical System”. Granted, that might not be the best name one could come up with. I actually thought that “cyber-” had disappeared in the 90s, but apparently it has come back. Pretty much every plenary speaker struggled to offer his definition of what a CPS is, but none was very convincing.

So let me maybe say here what CPS are in my opinion. It turns out that many researchers working in different areas, such as real-time systems, control systems, or sensor networks, face the same kind of problems and work on the same systems, but used to go to different conferences and publish in different journals. The goal of CPS week is to change that and foster collaboration, which I think is a good idea. But you need a new word to be able to bring these people together. Clearly, you cannot suddenly tell researchers in real-time systems that now they are doing control theory, and vice-versa. That’s why you can pretty much put whatever you want in your definition of CPS, depending on your background.

I know I haven’t posted anything on this blog for ages, but the past few months have been pretty hectic. Among other things, I’ll be on the job market this year, and I’m teaching a course on discrete-time stochastic control at Penn this term. All this is fun but time consuming. Maybe I’ll have more time for this blog after the HSCC submission deadline…