Analysis and framework-based design of a fault-tolerant web information system for m-health

Abstract

The e-health domain has the objective to assist and manage citizens' health. It concerns many actors like patient, doctors, hospitals and administration. Current and forthcoming generations of application will be web based and will integrate more and more mobile devices. In such application domain, called m-health, dependability is a key notion. In addition, more and more functionalities of such systems will be implemented as services for providing qualities like adaptability and maintainability. This paper presents, through a case study, how we can analyse and design an application that controls the insulin injection and that is embedded in a mobile device belonging to an e-health Web Information System (WIS). In order to ensure the dependability of the control systems, we show how to use Coordinated Atomic Actions (CAA). CAAs provide well defined concepts for fault tolerance, error-detection and error recovery in a distributed context where competitive and cooperative concurrencies are considered. The combination of CAA and SOA at design level it is proposed, with the purpose of being able to design systems that are partially implemented using service-oriented technologies. We updated and used our implementation framework, called CAA-DRIP, which originally was not tailored for service-oriented mobile applications. Thus, in this paper, we also propose an adaptation of CAA-DRIP for mobile devices.

Extended Fault Taxonomy of SOA-Based Systems

“…Belli et al [18] have shown fault taxonomy on the basis of sequencing pattern of the graph-based approach; positive sequencing faults, negative sequencing faults. Balbastro et al [19] have focused on the error during service delivery as a latent error and dormant fault.…”

“…Transaction isolation fault [19] Synchronized transactional fault tolerant approach Transaction evolving from a flat transaction, transactions with savepoints, nested transactions -A mechanism for handling exceptions in a synchronized way -FT (Fault Tree) technique for providing correct service in case of long-lived transaction Service unavailability fault [27], [24] Modelling approach -Temporal unavailability of service = Estimated execution time < end available time -Temporal service unavailability -Analysis impact region and applying repair strategies: replacement, re-composition, and renegotiation -Calculates temporal negative impact, inconsistent and satisfied OR consistent and unsatisfied OR inconsistent and unsatisfied.…”

“…Some transactions may be long-lived transactions that exploit the resources and can hold locks on external objects for long periods of time due to the infinite execution in the faulty situation. So, Balbastro et al [19] have discussed transaction isolation fault and proposed a mechanism for handling exceptions in a synchronized way.…”

“…Many researchers have primarily been focused on service delivery fault. Balbastrol et al [19] have developed CAA-DRIP framework as fault tolerant system to deal with latent errors and dormant faults and errors at service delivery in SOA. They have used simulation technique as repair assumption.…”

Fault Repairing Strategy Selector for ServiceOriented Architecture

“…SOA specific fault can be categorized as publishing fault, discovering fault, composition fault, binding fault and execution fault according the SOA life cycle stages as proposed by [6]. Service unavailability fault [7] [8], byzantine fault [9] [10], prescribed policy violation [11], timeout exception [12], SLA (Service Level Agreement) claim fault [13], latent errors and dormant faults [14], adaptation faults [15], interaction faults [16], network traffic [17] of service-oriented computing faults are noted to be highly emphasized faults by the researchers.…”