Context: Adaptive monitoring is a method used in a variety of domains for responding to changing conditions. It has been applied in different ways, from monitoring systems’ customization to re-composition, in different application domains. However, to the best of our knowledge, there are no studies analyzing how adaptive monitoring differs or resembles among the existing approaches. Objective: To characterize the current state of the art on adaptive monitoring, specifically to: a) identify the main concepts in the adaptive monitoring topic; b) determine the demographic characteristics of the studies published in this topic; c) identify how adaptive monitoring is conducted and evaluated by the different approaches; d) identify patterns in the approaches supporting adaptive monitoring. Method: We have conducted a systematic mapping study of adaptive monitoring approaches following recommended practices. We have applied automatic search and snowballing sampling on different sources and used rigorous selection criteria to retrieve the final set of papers. Moreover, we have used an existing qualitative analysis method for extracting relevant data from studies. Finally, we have applied data mining techniques for identifying patterns in the solutions. Results: We have evaluated 110 studies organized in 81 approaches that support adaptive monitoring. By analyzing them, we have: (1) surveyed related terms and definitions of adaptive monitoring and proposed a generic one; (2) visualized studies’ demographic data and arranged the studies into approaches; (3) characterized the main approaches’ contributions; (4) determined how approaches conduct the adaptation process and evaluate their solutions. Conclusions: This cross-domain overview of the current state of the art on adaptive monitoring may be a solid and comprehensive baseline for researchers and practitioners in the field. Especially, it may help in identifying opportunities of research; for instance, the need of proposing generic and flexible software engineering solutions for supporting adaptive monitoring in a variety of systems.

Introduction

Over the years, methods and techniques for monitoring a variety of systems have been proposed. There are approaches proposed for monitoring communication networks (e.g., Liu et al. [1]), buildings’ or persons’ health (e.g., Kijewski-Correa et al. [2] and Mshali et al. [3], respectively), software systems (e.g., Toueir et al. [4]), environmental conditions (e.g., Alippi et al. [5]), etc. Monitoring allows systems’ stakeholders checking how their systems progress or behave under different conditions, and reporting on relevant changes. However, it is often expensive and intrusive. Thus, the design of a monitoring system (i.e., the software system that implements monitoring capabilities) usually involves tradeoffs between the impact caused by the action of monitoring and its expected quality of results, such as data accuracy, freshness and coverage, among others [6,7]. In addition, a monitoring system is exposed to a diversity of runtime events, e.g., structural or operational changes on the System under Monitoring (SuM), faults on the monitoring system’s elements or the emergence of new monitoring requirements. In order to deal with all these challenging factors, software engineers have proposed different approaches for making current monitoring systems adaptive. Proposals have emerged from a variety of research fields (e.g., sensor networks, instrumentation, requirements monitoring). However, although these diverse proposals share most high-level challenges, solutions have been developed, evolved and kept isolated in those different fields. This hinders the discovery of synergies among the different proposals to adaptive monitoring as well as the standardization of the main field concepts, starting with the adaptive monitoring term itself, and the normalization of the challenges faced. To the best of our knowledge, there is not any work reviewing adaptive monitoring approaches across different fields. Thus, this work aims at uncovering and characterizing existing approaches supporting the adaptation of monitoring systems, in general. In order to achieve this goal, we have conducted a systematic mapping study (SMS) for identifying the primary studies on adaptive monitoring published in academic venues. We have retrieved and selected the studies conducting a rigorous protocol, defined in this work, which follows the guidelines presented by Petersen et al. [8] and Kitchenham & Charters [9]. For analyzing the identified studies, we have designed 5 high-level research questions (RQs) which we have divided into a total of 18 research sub-questions. To extract data from these studies, we have used a qualitative analysis approach based on the method describe by Miles et al. [10]. After the qualitative analysis, we have applied data mining over the extracted data for identifying patterns in the approaches. Concretely, we have used the rule-based algorithm JRip implemented by the data mining tool Weka [11].