Page 57 - Trust in ICT 2017

Trust in ICT 1

Various performance metrics that have been used to evaluate trust management schemes for MANETs. Note
that a single work may use multiple performance metrics. Standard system performance metrics typically
used to evaluate trust management systems; these metrics include overhead (e.g., control packet
overheads), throughput, packet dropping rate, and delay. “Route usage” refers to the number of routes
selected particularly when the purpose is for secure routing. “Trust level” is a recently used system metric.
Example metrics using the trust level include confidence level of the trust value, trustworthiness, opinion
values about other nodes, and trust level per session. “Others” indicates metrics that consider system
tolerance based on incorrect reputation threshold, availability, convergence time to reach steady state in
trustworthiness of all participating nodes, and percentage of malicious nodes.

6.2.8 Decision Making
Trust is collected to judge the credibility of the cooperative entities in the system, based on which make a
decision to deliver a service or application. The decision making is personalized, service/app-specific and
context-aware that is similar as trust. A machine learning mechanism should be used for decision making
trust provisioning in which all trust score, context, and user preferences are taken into account for making
good decisions.

6.3 Trust Provisioning in Networking Domain

6.3.1 Security and Privacy
Trust Establishment provisioning for security and privacy:

As mentioned before, Laih [26] proposed a challenge response protocol to identify malicious or unreliable
peers in P2P systems. The proposed protocol verifies every contacted peer and records the corresponding
trust value making it more effective than the traditional polling algorithms. Only in the worst case, the
protocol may use the same number of messages as a polling algorithm when the requesting peer specifies
the same Time to Live (TTL) and every peer returns all of its neighbours as referrals. Additionally, since all
challenge information is chosen at random, malicious peers have little opportunity to tamper with the P2P
systems. This protocol illustrates the details in the processes for rating, gathering, and trust construction. It
can be applied in both hybrid and distributed P2P networks.
Opposed to P2P networks, in open Multi Agent Systems (MASs), agents are owned by a variety of
stakeholders and they can participate or leave a system dynamically. It may be noted that participating agents
are likely to be unreliable, self-interested and possessed with incomplete knowledge. Moreover, since agents
are designed to behave intelligently and work in team therefore their intensions don’t remain static and
hence might change with time. Hence it is required to implement a protocol that could establish a level of
trust among interacting agents. In order to meet the above stated need, a trust establishment protocol has
been proposed in [19] by using existing protocol called contract net protocol (CNP) to help monitoring and
selecting their interaction partners.
6.3.2 Region

A trust management provisioning strategy for data usage policy in smart cities could be integrated with Smart
city Data manager for data analytic and data protection

A general architecture for Smart Cities consists of three layers:
• Infrastructure Layer: The layer contains variety of IoT objects that are deployed to send their data
to different applications. Because of IoT scenario, it considers that these IoT objects can belong to
different domains, such as, smart sensors from the WSN domain, smart street lights/traffic signal
poles from smart city domain or home alarms system/intelligent Heating, Ventilating, and Air
Conditioning HVAC system from smart home/building domain. It also considers that some kind of
infrastructure access/control mechanism is used by each of these domains independent of each
other’s.
• Platform Layer: The layer consists of the several functional entities: Trust Manager, Ontology
Manager, Policy Manager, Data Manager, and Application Manager. For the trusted data usage
model, the Trust Manager will collaborate with the Ontology Manager and Data Manager to set the

49