Page 192 - Trust in ICT 2017

3 Trust in ICT

7.4.1 Trustworthy data collection and aggregation

As the number of data sources and types are dramatically increased, the trustworthiness of data itself is
regarded as important. Because collection and aggregation of false data will lead to a degradation of service
quality and waste of system resources, it is a significant issue to detect wrong or polluted data. Trust metrics
and models can be used as criteria for checking trustworthiness to achieve trusted data collection and
aggregation.

7.4.2 Trustworthy data process and analysis

When the huge amounts of data are collected to a system, these data should be processed and analysed in
trustworthy ways. Data process and analysis mainly occurs in cyber domain (e.g., utilizing cloud computing
for big data analysis), however, it also can be done in a physical domain as well as a social domain. Each
domain has its own intelligence to process incoming data to create new useful information. This information
is usually propagated to different entities and domains, so there are some ways to check whether given data
process and analysis mechanism is trustworthy or not. Measurable trust value should be defined to analyse
trust of entities, and it is also important to find appropriate trust evaluation mechanisms for analysing trust
values for a specific domain.

7.4.3 Trust metric and modelling
A trust metric is a measure to evaluate a level of trust by which a human or an object can be judged or
decided from trustworthiness. It can be differently defined in each human or each object. Trust metrics might
be separately defined in each of domains, but the key issue is to describe qualitative and quantitative metrics
across the domains, to determine the attributes in the different domains. For measurable trust, some
mechanisms and solutions may be established by defining trust metric. There are several attributes social-
cyber-physical domains for trust provisioning. Attributes in each domain of Figure 6-1 are examples.
Depending on the services and applications, the required attributes of trust may vary.

A trust model is the method to specify, build, evaluate and ensure trust relationships among entities. The
trust model is used for the processing trust data. Most existing trust models are based on the understanding
of trust characteristics, accounting for factors influencing trust. Trust modelling is domain-specific and there
exists numerous ways to define trust model for each domain. It is a critical issue to select a suitable trust
model for a particular domain.
7.4.4 Trust index

A trust index is a composite and relative value that combines multiple trust related indicators (e.g., objective
trust metrics and subjective trust attributes) into one benchmark measure, which is similar to ICT
Development Index (IDI) or stock market index. It can be used to compare trust among stakeholders when
they create a new trust relationships or a trust value chain. The trust index should be designed to quantify a
trust value of each stakeholder, and the methodology used to compute trust index should be clearly defined.
In order to apply the trust index to a real world, common indicators for covering different stakeholder
characteristics and comparing methods for trust indices of different stakeholders should be developed.

7.4.5 Dissemination of trust information
Trust dissemination means to distribute or broadcast trust information. There could be many ways of
disseminating trust information in different domains. In case of a social domain, recommendation and
visualization methods are considered as main approaches to disseminate trust information [b-Sherchan]. The
efficient, effective and suitable trust dissemination methods should be developed.

7.4.6 Trustworthy system lifecycle management
In order to achieve trustworthy systems, we need a systematic methodology to cover all relevant trust
aspects of operation life cycle. At the design phase, the definition, metrics and goals of trust for the target
system should be determined and the system should be developed while trust measures are considered to
fulfil the design goals in the development phase. The maintenance phase has to properly monitor the normal
operation of the running of a trustworthy system and the dynamics of the execution environment to verify
the trust provisions at runtime.

184