In environments where services can be dynamically composed, pre‑configuration of accounting components is no longer a viable approach. Service compositions are created and executed within a short time span, so there is no time for manual configuration of appropriate accounting operations. Business logic for accounting (“accounting logic”) must be automatically configured when service compositions are initially constructed, or subsequently modified. In addition, this accounting logic will need to adhere to the potentially complex business agreements between the providers of the services that comprise a composed service. These observations reflect the central question to be addressed in the proposed research programme:

How must accounting systems evolve to provide effective and efficient accounting for dynamically composed services?

Accounting systems embody complex functionality, ranging from the collection of service usage data to invoicing customers. A programme of this size cannot provide a complete solution, applicable to all application domains, for accounting for dynamically composed services. However, by limiting its scope to a subset of accounting functionality in a particular domain, it can develop targeted solutions, from which general principles for supporting accounting for dynamically composed services can be extrapolated.

The programme will thus limit its scope to mediation, usage-based charging and rating aspects of accounting for dynamically composed ubiquitous communications services. Ubiquitous communications services are chosen since their user‑centric focus makes them ideal for demonstrating the commercial potential of dynamic service composition. Similarly, from an accounting perspective, charging schemas and their application to mediated usage records in rating engines is the core of the entire process, since it is here that the business model of the service provider(s) is effectively translated into charges subsequently relayed to customers. Conversely, it should be highlighted that techniques for dynamic service composition are outside the scope – the programme will be limited to addressing how composed services are accounted for after they have been constructed by a service composition engine.

Within the identified scope of the work programme the major issues listed below will be addressed:

• How are usage data for individual services
  identified as relating to a particular invocation of a dynamically composed service? How can mediation components be automatically configured to do this correlation?
• How can complex business relationships between
  service providers be manifested in mediation and rating accounting logic in a manner that is amenable to the non-technical user?
• How can we automatically generate a charging schema for a dynamically composed service that satisfies business requirements?
  How can generated charging schemas be automatically deployed in rating engines?
• How can we increase the degree of automation
  involved in configuring accounting components without compromising their efficacy, performance and reliability?
• If a composed service is modified during its execution will certain changes necessitate re‑rating of usage data?
• What are the implications of faults in the execution of composite services? (for example, are charges generated for constituent services already successfully executed?)

Solutions to these will constitute the main outputs of the programme, which will collectively answer the central question of how to evolve accounting systems to support dynamically composed services. As illustrated in Figure 1 below, a domain-specific language for accounting logic, coupled with tools for automated transformation into platform-specific accounting logic, will be used to realise automated configuration of accounting components for dynamically composed services. A central role is played by the Accounting Logic Generator (ALG), a new type of component to be developed during the work programme. The approach can be described in terms of six steps:

                              i.    Composition:
The Service Composer constructs a composite service to satisfy user requirements. This step is outside the scope of the proposed work programme;

                            ii.    Transfer:
Accounting logic models relating to all the constituent services of the composed service and composite service identification information is transferred to the ALG. For rating, services are envisaged as being associated with a dual level charging schema model, in which the lower level details how a service is charged when invoked in isolation, and the upper level dictates how these charges change if the service is used in conjunction with other services;

                          iii.    Consolidation:
The ALG analyses the accounting logic models of the individual services and consolidates them into a single model for the composed service. The consolidated model will reflect any amendments to charging schema that must be made due to business relationships between the providers of the services that make up the composed service;

                          iv.    Transformation:
Mapping languages are used to transform the consolidated accounting logic model into platform-specific mediation and rating code. This approach allows for the manipulation of service accounting logic models at a high-level of abstraction, whilst ensuring that deployed software uses the speed and reliability optimised languages used by accounting system vendors;

                            v.    Deployment:
The ALG then deploys the generated platform-specific logic on the meditation and rating components. The programme will specify protocols and component interfaces that allow efficient, reliable and secure deployment of accounting logic;

                          vi.    Collection:
Once the accounting logic has been deployed the composed service can be executed under control of a workflow manager (not shown). The mediation component collects metering records relating to the service and processes them according to the deployed logic;

Based on the above we can identify the main outputs of the programme as follows:

1. A domain-specific modelling language for accounting logic;
   
2. Models of accounting logic for ubiquitous communications services, including dual level charging schema models;
   
3. Algorithms for the consolidation of accounting logic relating to constituent services of a (dynamically) composed service;
   
4. Mapping languages for the transformation of accounting logic models to mediation and rating code;
   
5. Protocols for dynamic deployment of mediation and rating code;
   
6. Interfaces for deployment of platform specific accounting component code;
   
7. Specification of enhanced mediation/rating processes for dynamically composed services.

Figure 1: Accounting Logic Generation and Deployment

Throughout the programme all research outputs (languages, models, algorithms, protocols, interfaces and processes) will be rigorously validated to ensure their efficacy, usefulness and efficiency. Validation will take place via experimental trials carried out using prototypes deployed in a test bed environment. Two types of experiments will be carried out: quantitative experiments that will verify the functional efficacy of the developed artefacts and assess their performance against defined criteria; and qualitative trials in which expert “test pilots” will assess the degree to which the developed artefacts are likely to improve the process of configuring accounting logic.

Summary of outputs

Refereed Academic papers

Jennings, B., Malone, P., & van der Meer, S. 2005, "A Two Phase Rating Process for Dynamically Composed Services", in Proc. 12th Workshop of the HP OpenView University Association (HPOVUA2005), B. F. Marques, T. Nebe, & B. F. Oliveira, eds., pp. 155-171.

Jennings, B., Malone, P., & Gaughan, G. 2005, "Charging for Dynamically Composed Services in the Digital Business Ecosystem", accepted for publication in Proc. eChallenges 2005.

Non-refereed academic papers

Gaughan, G. & Jennings, B. 2005, "An Aglorithm for Two Phase Rating of Dynamically Composed Services", in 9th Institutes of Technology Science & Computing Research Colloquium (ITSCRC 2005).

Gaughan, G. & Jennings, B. 2005, "Two Phase Rating of Dynamically Composed Services”, accepted for publication in Proc. 2005 Information Technology & Telecommunications Conference (IT&T 2005).