C ausas de las alteraciones psicosomáticas

Description

UbiREAL is a ubicomp environment simulator or smartspace simulator as the authors refer to it, developed by Nishikawa et al. [80]. UbiREAL incorporates simulators for physical quantities, network simulation and a 3D visualisation mechanism. This enables the simulator to facilitate deployment of virtual devices in a 3D space, simulation of communication between devices from MAC level to application level and to simulate changes in the physical state of the environment, e.g. temperature change. The focus of UbiREAL’s evaluations are invisible physical quantities such as temperature, humidity, electricity and radio as well as audible quantities such as acoustic volume.

The most relevant aspect of UbiREAL with respect to this thesis, is the approach towards systematic testing of the correctness of a ubicomp system. The authors have approached this challenge by providing a formal model for system specification. A smartspaceU is defined as a tupleU = (R, D), whereRis the set of rooms andD is the set of devices. This is similar to the set-based location models mentioned earlier in this chapter. The service specification is defined using a set of rulesAP =l1, ..., ll,

where each rule l is comprised of a condition and an action. The example service specification that the authors provide is an air-conditioning system. If a room exists at a certain temperature that is defined as hot by a user, then switch on the air-conditioning unit.

To test an implementation of the system specification, a tester must either manually specify routes for mobile objects through the smartspace, or define a test specification which contains a sequence of test cases or scenarios that are automatically executed. These sequences are tested against a set of propositions,P, which if they hold true, infers that the system is correct. The authors fundamentally aim to determine correctness of a system specification by testing its implementation to determine if the following conditions hold:

1. For every rule l = (c, a)∈AP and every state s of U, if condition cholds for state s, then actiona is executed.

2. Every proposition p∈P holds.

Due to the large state spaces that often must be tested for ubicomp systems, the authors advocate testing either using a predefined set of samples or at predefined time intervals.

The authors present performance results for sequence execution and simulation visualisation. The authors do not appear to have published example output or results from a defined set of propositions. The authors also do not appear to have published performance results for test execution.

Analysis

The major contribution of this work is an approach for formal specification of smartspace applications, along with an approach for testing correctness of these specifications. Additional strengths include the extensive simulation capabilities that feature in the work, both in terms of physical environment phenomena and network connectivity between devices. Also worth noting is that UbiREAL is one of few tools which attempt to address a full prototyping cycle, from system specification through to testing.

However, the use of a set-based location model is a limiting factor in terms of the reasoning that can be conducted about the spatial relations of the ubicomp

environment. Moreover, UbiREAL as a tool suffers from a lack of generality. The service specification must be defined in specific terms. The examples provided by the authors refer to individual users and devices. It is also difficult to assess the expressiveness of the tool because the user, device and room models are not formally documented in order to illustrate their attributes.

In terms of testing, a limitation of the work, which the authors acknowledge, is that they can only validate correctness for the inputs that are known to have been tested. This is because of the challenges surrounding the scale of testing required in order to address the full state space. The authors advocate eliminating test-cases that are unlikely to occur, however adopting an approach such as this does not fit well for testers who may wish to investigate unanticipated and outlier test cases. Additionally, because the inputs are tested against a system specification, the tests are effectively locked into the assumptions made in the system design. This leaves less room and flexibility to investigate the key challenges of unanticipated behaviour and unexpected situations.

The authors refer to the use of temporal logic, such as Computational Tree Logic, which they describe as being used to intuitively determine the state to which a system will transit. However, the details on how this is used to investigate correctness are insufficient to draw conclusions about the contribution of this particular aspect of the work. The authors also claim that inconsistencies in system specifications should be detected by UbiREAL, however the description of this is short and lacking detail. It appears that detection of these inconsistencies is through observation and visual monitoring of the simulated smart space by the tester.

Overall, the contribution of this work is notable since few tools have addressed the issue of correctness and validation for ubiquitous computing systems and smart spaces.