wewst11 trustworthiness monitoring of dynamic service compositions v2
TRANSCRIPT
Trustworthiness Monitoring of Dynamic
Service Compositions
Hisain Elshaafi, Dmitri Botvich, Jimmy McGibney
Outline
• Trust and trustworthiness.
• Aniketos – secure and trustworthy composite
services.
• Operation of service (re)composition and
trustworthiness monitoring.
• Trustworthiness prediction and aggregation.
• Experiments.
• Conclusion and future work.
Trust and Trustworthiness
• Trust:
– a relationship between two or more entities
– indicates the contextual expectations from an entity
towards another
– regarding accomplishing an action at a certain quality.
• Trustworthiness (TW):
– the level of trust that the trusting entity has in another.
• Reputation:
– the information available about an entity from various
sources
– such as QoS and user ratings
– can be used to determine its TW.
Trustworthy Composite Services
• Composition techniques must be able to identify which
component Web Services (WSs) are trustworthy:
– helps consumer confidence, and
– provides a safe environment for businesses to dynamically
interact and carry out transactions
• The composition techniques must be able to maintain
the most trustworthy and cost efficient Composite
Service (CS).
• Therefore, addressing trust is essential for the success
and adoption of the services paradigm.
• FP7 Integrated Project
• Sept 2010 – Feb 2014
• Aniketos addresses establishing and maintaining
trustworthy and secure service compositions during
design and run-time.
• Including runtime monitoring and adaptation of services.
• The adaptation is needed due to:
– changing operational, business or threat environments or
– changes in service quality and behaviour.
Operation of WS (Re)composition
• A component WS that is below the satisfactory TW level can
be replaced with another WS offering the same capability but
with better TW.
• The Cost Engine
determines the cost of the
CS as a result of the
change.
• This ensures that a
balance is maintained
between both TW and
cost efficiency of the WS.
Trustworthiness Monitoring
• Incoming events are evaluated by a rules engine to
generate WS ratings.
• The rules calculate the
rating for the event and
add other attributes
including the time-stamp
and the type of event.
• Ratings are stored and
can be used by the trust
engine to calculate the
overall TW level of the
WS.
Trustworthiness of a Composition
• The calculation of TW level depends on the
structure of the business process.
Trustworthiness Aggregation
WS Trustworthiness Prediction
• Trustworthiness level is determined by two
values; trust score and confidence score.
• Trust score is a dynamically weighted moving
average of the rating scores.
• The weighting is based on the recency of the
ratings.
WS Trustworthiness Prediction (cont.)
• Confidence score is calculated as a product of quantity
confidence and quality confidence:
– Quantity confidence indicates the number and the recency of
ratings on which trust score is based.
– Quality confidence indicates the stability of the ratings values;
frequent fluctuations result in low confidence.
Experiments
• Decline of trust score of component service s1 leads to
decline of CS trust score.
Experiments (cont.)
• Decline of confidence score of component service s1
leads to decline of CS trust score
Experiments (cont.)
• Decline of trust (or confidence) score in different types of
constructs:
Loop Exclusive Choice
Experiments (cont.)
• Decline of trust (or confidence) score in different types of
constructs:
Discriminator
Experiments (cont.)
• Performance comparison to FIRE algorithm:
Conclusion and Future Work
• Approach to maintaining TW of dynamic composite
services.
• Calculation of TW and cost depends on the construction
of the composite service.
• Develop optimisation mechanisms and algorithms for
dynamic WS composition and adaptation.
• Taking into account multiple objectives including TW,
cost and pricing.
• Maintaining trustworthiness through resource
management
