Autonomic Aspects of IoT Based Systems: A Logistics Domain Scheduling Example - Interoperability and Open-Source Solutions for the Internet of Things - page 160

insert into CrossroadTrafficJams select ∗ from
CrossroadCarsNumber (numberOfCars > TRAFFIC JAM THRESHOLD

1

Listing 1.3. Detect trac jams.

4.4 MAPE-K Loop Plan Step

The planning step in our solution relies on CS techniques, in fact the application
of CS for scheduling problems. The decision of this approach is natural since
scheduling represents the problem of fulfilling a set of soft and hard constraints
involving the temporal dimension as well.
Constraints Satisfaction Problems (CSPs) are meant to identify a set of valid
values for variables which satisfy the requirements defined as constraints [ 19 ].
A CSP is defined using a {V,D,C} tuple where [ 20 ]:

V = {v 1 ,v 2 , ..., v n }, (1)

D = {D 1 ,D 2 , ..., D n }, (2)

C = {c 1 ,c 2 , ..., v m }. (3)

V represents the set of variables, D defines contains the variable domains

and C is the set of constraints. A constraint c k is defined as a cartesian product

of the variable domains D 1 × D 2 × ... × D n .

Different techniques and algorithms have been developed and implemented
for CSP. The solution proposed in this paper is based on the Choco solver [ 9 ].
In our case, the planning step consists of scheduling the minivan agents'
delivery plan based on the possible routes modeled as a graph an a set of hard
an soft constraints. The hard constraints define the items which have to be
delivered and their corresponding recipients, while the soft constraints define
client time windows (when a delivery is possible) as well as minimization of the
costs.
Based on the inputs obtained after the analysis step about the tra c condi-
tions and the delivery requirements specific to each constructing site, the vari-
ables and the constraints of the problem are defined and fed to the solver. The
expected outcome is a set of possible routes. The method is described in detail
in [ 21 ].

4.5 MAPE-K Loop Execute Step

In order to close the loop, after the system has detected that a trac jam has
occurred (see Sect. 4.3 ) and the new routes have been generated (see Sect. 4.4 ),
it sends the new routes to the minivan agents.
As mentioned above, the behaviour of the agents, from the CoReMo simu-
lator platform, is described using PNs. In Fig. 6 is presented the the PN which