[Thesis notes] Quality-of-Service Oriented Web service composition algorithm and Planning Architecture (JSS,

Time: 2.8 hours
Jong Myung Ko, Chang Ouk Kim, ick-hyun Kwon.Quality-of-Service Oriented Web service composition algorithm and Planning Architecture. Journal of systems and software, vol 81 (11), Nov 2008, pages 2079-2090 (GS: 3)

This paper mainly discusses the issue of QoS aware service composition. Although it was published in, the submission date was as early. so far, this problem has been "Study extensively", and it is unlikely that there will be another major innovation. the following are the thesis notes.

1. This article mainly proposes a solution to the QoS aware service composition problem.AlgorithmCalled "Constraint Satisfaction Based Web service composition algorithm" (s2.4). This algorithm uses some features of tabu search and simulated annealing "meta-heuristics. then, an implementation framework named "QoS-Oriented Web service composition planning custom uture" is provided ".
In table 3, the core algorithm is not complex. It is an empirical algorithm, mainly including three points:
(1) how to generate an initial plan? (S2.4.1)
(2) how to generate "neighbor plan "? (S2.4.2)
(3) how to determine whether "neighbor plan" is used "? [Exp {f (S)-f (s) * iteration}> random (0, 1)]

* Tabu search (Taboo Search)
First proposed by Glover (1986), "an extension of local domain search is a global gradual optimization algorithm that simulates human intellectual processes ". "the most important idea of Taboo Search is to mark some objects corresponding to the local optimal solution that has been searched, and try to avoid these objects (rather than absolute prohibiting loops) in further iterative search ), this ensures the exploration of different effective search methods. "(See Baidu encyclopedia)

2. This Article involves six Qos attributes (s2.1): execution cost, execution time, availability, successful execution rate, reputation, and frequency.
Availability is defined as "the ratio of the time that Web service is available for immediate use. It is measured as AV = uptime/(uptime + downtime )."
Successful execution time is defined as "counted as the number of successful executions divided by the total number of service trials ".

3. (S3) introduced "Planning Architecture". The author just proposed this framework and did not implement this system.
(S4) is the experiment part. Compare the algorithm (two initial planning methods, reputation & frequency based, local optimization) with the IP address method of zeng04. it is strange that, when using the IP method, it takes tens of minutes to solve dozens of atomic processes and dozens of alternative services. Is it so exaggerated?