As the Future Internet arrives, more complex, service-based applications are spreading. These applications pose several challenges, including the huge amount of entities that must interact and their heterogeneity. The success of these applications depends on the collaboration and communication of these entities, that might belong to di erent organizations and administrative domains. Therefore, trust and reputation become two crucial issues. We propose the speci cation and design of a service-based security architecture that stresses the delivery of trust and reputation services to any application that might require them.
architectures can be derived. However, architects of service-oriented systems
often nd con icts when trying to reconcile the business goals, the non-functional
requirements, and some principles that are a consequence of the SOA approach,
such as standardization at multiple levels, loose coupling, reusability,
composability, and discoverability [
A crucial component for many SOAs is the Enterprise Service Bus (ESB). An ESB is a complex piece of software that mediates between clients and services, and among services. It is an information bus that connects and allows the communication of heterogenous applications (e.g. Java Message Service (JMS) applications and Simple Object Access Protocol (SOAP) applications). It also caters for other commonly needed services, such as security, protocol conversion or exception handling. 1.2
Trust and Reputation in Service-Oriented Architectures for Future Internet Applications SOA systems have been traditionally secured at the endpoints. This means that a service is hard-coded with security functions that realize security services. This approach has several shortcomings. One of them is the business and security coupling, which in turn makes manageability and interoperability more di cult. Reusability is also undermined, since although the service might ful l a business goal interesting for the organization, the coupling with security would make it useless for the organization if the security policy of the service does not match that of the organization.
The Future Internet entails the arrival of new, complex service-based
applications that span across multiple boundaries and administrative domains. This
calls for trust and reputation services that assure the trustworthiness of the
entities that take part of such heterogenous communications and collaborations,
and which aid the decision-making processes. Examples of applications are those
described in NESSoS [
Trust and reputation services might assist traditional security services, such as encryption and authorization. The latter mechanisms can provide a trusted environment, where communication between service providers and clients is protected, and where access to business assets is limited. These mechanisms, if well leveraged, might provide certain degree of trust between clients and customers in general. However, the notions of trust and reputation focus on speci c relationships between a given provider and a client, taking into account both local information (e.g. interaction records), and external information (e.g. recommendations from other clients or providers). Thus, traditional security mechanisms should provide the basis, the trusted medium onto which to build trust and reputation solutions.
Although there are some proposals towards the delivery of security as
services, to the best of our knowledge none of them cope with trust and
reputation services. Hafner et al. [
As we mentioned in Section 1.2, the arrival of Future Internet demands trust and reputation solutions. In the complex, heterogeneous scenarios that arise in this context, trust mechanims must be leveraged since they play a crucial role in decision-making processes.
Trust is a complex concept for which a clear, standard de nition has not been
provided yet. A possible de nition is the level of con dence that an entity of a
system places on another entity of the same system for performing a given task.
Thus, two features of trust are uncertainty and subjectivity. Reputation, on the
other hand, is a more objective concept, and is based on information about or
observations of the past behaviour of an entity. Both concepts are very related,
and in fact, reputation can be used as a means to determine whether an entity
can trust another entity [
Our aim is the speci cation and design of a functional, reusable servicebased security architecture that emphasizes the delivery of trust and reputation services.
A key feature of this architecture must be its reusability, since it should serve as a framework for building multiple applications that require trust and reputation services, as it is the case of Future Internet applications. It is also important to note that we intend to analyze the relationship between trust, reputation, and other security services. The architecture might therefore deliver those security services that have an in uence on the overall trust of any application.
In the literature there are important contributions in the eld of trust and reputation that might assist our work.
Kiefhaber et al. [
It is also important for a reusable architecture to o er exible mechanisms
to accommodate or compose di erent trust models according to the needs of
the application. In this direction, Huynh [
The methodology that we intend to follow to conduct the research consists of the following phases: 1. Exploratory phase: during this phase, a wide research will be done in order to gain a solid knowledge about the process and tools for architecting a service-oriented architecture. 2. Construction phase: the speci cation and design of the architecture will be done during this phase. In order to accomplish these tasks, we will accommodate the output of the previous phase, which provides us with a set of tools, methods and processes that currently exist to architect service-oriented solutions. This output could contain as well some found gaps that we will need to bridge during this phase. 3. Validation phase: NESSoS scenarios represent some of the most important Future Internet applications. Thus, the architecture will be validated for, at least, one of these scenarios, which are within the scope of e-Health or Smartgrids.
At the moment, we are at the exploratory phase, where we are surveying existing approaches towards the building of a service-based security architecture. Some of our ndings are brie y described in the next section. 3.1
Processes and Tools for Architecting SOA Solutions
In general, the process for building any architecture is iterative and consists
of several steps [
This generic process is also applicable to the case of SOA. Yet there are
some more speci c SOA-oriented processes, as the Rational Uni ed Process for
Service-Oriented Modeling and Architecture (RUP SOMA) proposed by IBM
[
The modeling of a service-oriented architecture requires designing many
aspects, ranging from the business activities to the non service-oriented, existing
assets. However, the most interesting for us is the service model, where both
the atomic and composite services are de ned in terms of their speci cations,
interfaces, inputs, outputs, and collaborations. As with traditional object or
component-based applications, services can be graphically modeled with the
Uni ed Modelling Language (UML). Several UML pro les exist in the
literature that allow modeling service-oriented architectures, such as UML4SOA [
For the implementation and deployment of services, there are several
alternatives. One of them is Apache Tuscany [
As in the case of the processes, we will research on these tools in order to choose the one that better ful ls our needs. 4
New Future Internet applications will need support from trust and reputation services for their successful adoption. Whereas traditional security requirements have been provided by architecture middlewares and frameworks, trust and reputation have often been laid aside.
On the other hand, the service-oriented paradigm provides many advantages, including its focus on reusability. We propose the building of a service-oriented architecture, capable of delivering security services in general, and emphasizing the delivery of trust and reputation services in particular. This architecture must be exible and reusable, constituting a framework that can be used by di erent applications that might require their services.
The main contributions that we expect to do in the eld of Engineering Secure Software and Systems are in di erent levels. At a process level, we will research on the high-level development processes for building service-oriented solutions. At the architectural level, a research will be conducted regarding the modeling issues for service-based architectures. At a security level, the relationships between security services will be analyzed and made explicit, which in turn might help us to gain a better understanding of how trust is related to more traditional security services. Finally, from an overall perspective, we believe that the process of building a service-based security solution bridges a gap between the service engineering and security research communities.