Backtracking algorithms for service selection

TL;DR

This paper introduces backtracking algorithms for automating service selection by mapping it to a CSP, enabling efficient real-time optimization of service compositions under constraints.

Contribution

It presents a novel approach that maps service selection to CSP and develops backtracking algorithms inspired by CSP heuristics, improving efficiency over exhaustive search.

Findings

Backtracking algorithms outperform exhaustive search in runtime.

The approach finds optimal solutions in real-time for small and medium compositions.

Experimental results confirm the effectiveness of the proposed methods.

Abstract

In this paper, we explore the automation of services' compositions. We focus on the service selection problem. In the formulation that we consider, the problem's inputs are constituted by a behavioral composition whose abstract services must be bound to concrete ones. The objective is to find the binding that optimizes the {\it utility} of the composition under some services level agreements. We propose a complete solution. Firstly, we show that the service selection problem can be mapped onto a Constraint Satisfaction Problem (CSP). The benefit of this mapping is that the large know-how in the resolution of the CSP can be used for the service selection problem. Among the existing techniques for solving CSP, we consider the backtracking. Our second contribution is to propose various backtracking-based algorithms for the service selection problem. The proposed variants are inspired by existing heuristics for the CSP. We analyze the runtime gain of our framework over an intuitive resolution based on exhaustive search. Our last contribution is an experimental evaluation in which we demonstrate that there is an effective gain in using backtracking instead of some comparable approaches. The experiments also show that our proposal can be used for finding in real time, optimal solutions on small and medium services' compositions.