Where do I publish my research and why?
My work tends to be practical systems-oriented though often deriving from rigorous analytical underpinnings and accompanied by a mathematical analysis under some simplifying assumptions which brings out the trends in the protocol. I publish my work at the top system-level conferences and a few select journals. All our conferences (and of course journals) are archived. The conference papers are as readily available as the journals – electronically in the digital libraries of the respective publishers (ACM DL, IEEExplore, Elsevier, Usenix’s digital libraries) and in print in the libraries.
Among the conferences, the premier venue for all of dependability work is Dependable Systems and Networks (DSN). This used to be called the Fault Tolerant Computing Symposium (FTCS) and was started in 1970. It attracts all the top researchers in the field and has two tracks within it – the Dependable Computing and Communications Symposium (DCCS) and the Performance and Dependability Symposium (PDS). Broadly speaking, DCCS targets design of new dependable systems and PDS targets the validation and evaluation of such systems. The papers in DSN are 12 pages in length since 2011 (double column, 10 pt font, single space) (earlier they were 10 pages) and include detailed design, implementation, and results. The conference is attended by about 350 people and the papers are the most highly cited in the dependability literature. Each submitted paper is complete in every respect (i.e., acceptance is not based on just an abstract) and is reviewed in a two phase manner – in the first phase each paper is reviewed by 3 experts in the field (overwhelmingly Program Committee members) and papers that make it to the second phase, get an additional 2 reviews. All the papers are discussed in an in-person Program Committee meeting in which the merits and demerits of each paper are hotly debated and a paper is accepted under a high bar for intellectual novelty, solid design, and rigorous evaluation. The acceptance rate for DSN tends to be between 20-25% with the rate trending downward over the last 3 years.
The IEEE International Symposium on Reliable Distributed Systems (SRDS) is a venue that is more focused than DSN, has been around since 1980, and is considered a prestigious forum. It focuses on distributed software systems and the papers typically have a software protocol focus. It gets a fewer number of submissions than DSN in view of its focused nature. The reviewing and acceptance processes are similar to those in DSN with acceptance rate of between 20-30% and 20-30 accepted papers.
In addition to these venues, several systems conferences take papers related to dependable distributed systems. A sampling where I publish are given below. All these conferences take full papers (10 pages or more in camera ready form) and are rigorously refereed:
· ACM Conference on Embedded Networked Sensor Systems (SenSys): The premier venue for presenting work on embedded wireless networks that has a highly practical systems focus. So I strive to publish my work on the third research thrust of mine (Dependability of embedded wireless devices and networks) if it meets the practical systems-y requirement, in Sensys.
· IEEE/ACM Supercomputing Conference (SC): This is the premier venue for work on all different aspects of supercomputing, with a practical bent to it and a focus on evaluation at large scales or with realistic workloads. It historically has acceptance rates of 20-25%. Much of my work in the first research thrust (Dependability in large-scale applications) has been published here.
· ACM International Symposium on High Performance Parallel and Distributed Computing (HPDC): The conference is focused on high performance systems and I can submit my work in the first research thrust of mine (Dependability in large-scale applications) to this forum. It takes about 25 papers with acceptance rate of about 15-25%, with a surprisingly low rate in 2011. The conference is in its 21st offering in 2012.
· Security and Privacy for Emerging Areas in Communication Networks (Securecomm): The conference is new but appears to have good participation from solid security researchers in it. It started off as being focused on security in wireless networks but has broadened out beyond that into mainstream security topics as well. The conference takes about 30 papers with acceptance rate of 20-25%. The conference was started in 2005.
· IEEE Conference on Computer Communications (INFOCOM): This conference is considered the primary venue for networking work, both wireline and wireless, that has an analytical bent to it. However, for analytical networking work, appearance in a journal, such as IEEE/ACM Transactions on Networking, beyond an appearance in a conference, is considered the logical culmination for top quality work. I have published some of my work that has a mix of analytical and practical networking and even distributed systems context, in INFOCOM. I do not work in the purely analytical aspect of networking. The conference papers are 9 pages long, the reviewing process has an in-person Program Committee meeting and I have gotten 3-4 reviews for my submissions to INFOCOM.
What about core wireless networking conferences?
These are Mobicom and Mobihoc highly selective venues for publishing wireless protocol work. These are possible venues for my wireless dependability work but often I have opted in favor of sending such work to DSN because of the more targeted audience. Mobihoc has a more analytical flavor than my work is ideally suited for, with the rare of exception of our work in Mobihoc 09.
What about core security conferences?
These are CCS, Security & Privacy, and Usenix Security. These are definitely possible venues but there is overlap with DSN. Especially when the work has the flavor of tolerating natural faults, I consider DSN to be a better venue and also one where I have a greater familiarity with the community. It is my goal in the future to try and publish some of work in the second research thrust of mine (Strengthening enterprise-class distributed systems) in these core security conferences.
We have a dedicated journal for our area – IEEE Transactions on Dependable and Secure Computing (TDSC) since 2004. It was quarterly till 2010 and then it became bi-monthly (once every two months) in 2011. Each paper gets reviewed by 3 experts in the field, the time from the submission to the time a paper gets accepted (not published), assuming one major and one minor revision is about a year in my experience. The length of a paper is typically 14 pages, compared to 12 pages for DSN and 14 pages for some other systems conferences I publish in (Sensys, Usenix ATC), thus allowing for somewhat more expanded coverage in some cases.
There are other journals which also publish some dependability related work. My work on wireless dependability can be submitted to IEEE Transactions on Mobile Computing (TMC), ACM Transactions on Sensor Networks (TOSN), Elsevier Journal of Ad hoc Networks. My work on dependability in the wired systems that has a security angle can be submitted to IEEE Transactions on Parallel and Distributed Systems (TPDS), ACM Transactions on Information and System Security (TISSEC) and Elsevier Journal on Computer Networks. Overall, the time from submission to acceptance in my experience is from 12 months (for the more recent journals) to forever (if you are really unlucky with the Editor). In terms of reviews, the number of reviews is less than for the premier conferences and the quality of the reviews is more uneven than in the case of premier conferences, but average case is comparable to that of the conferences. The process of revising a paper has helped improve some of my journal papers. On the flip side, some of the work gets dated by the time it finally appears in the journal in print. Further, I find particularly valuable to my work the feedback I get from attendees at conferences and the discussions that follow my better-quality and better-presented papers.
Last modified: Sunday, June 03, 2012