You are on page 1of 16

Extending Traditional

Algorithms for Cyber-Physical


Systems
Sumeet Gujrati and Gurdip Singh
Kansas State University
Problem Statement
• Define mutual exclusion, predicate detection, and global state
recording problems over physical systems (PhyS), and implement
the solutions over a cyber-infrastructure (CyS) which monitors the
PhyS (thus forming a CPS), and maintains the state of PhyS.
• Define a spatial model of PhyS, and then define the model of CPS.
• Define an event model for CPS.
• Device techniques so that CyS maintains a consistent state of
PhyS, as inconsistencies may arise due to sensing delays, missing
sensor data, and limited sensing range of sensors.
• Device an approach to model user behaviors, as users are not
passive entities.
• Perform detailed simulation of proposed algorithms.
Related Work
• The field of CPS is inspired by WSN. The facts that the algorithm
must adapt itself depending on user’s behavior, and the algorithm
manipulating user’s behavior (by helping him) are missing.
• The problem of resource detection (e.g., detecting free parking
spaces [6]) has been widely studied, with their own limitations.
• Space models such as [3] are suitable for our work with few
modifications. Space and time aware predicates have been studies,
such as in [3,4,1], but they have their own shortcomings.
• [5] discusses constructing snapshots for a smart construction site. It
is more focused towards interpreting missing sensor data.
Proposed Idea
• We define an hierarchical spatial model of the PhyS.
• We define a graph based model of CPS.
• We define user behaviors in terms of their capabilities.
• We define an event model for CPS.
• We define mutual exclusion and predicate detection algorithms
(which are implemented in CyS) for different types of user behavior.
For example, a disciplined user will always follow CyS instructions,
whereas, a complete random user will act on his own.
• The algorithms we propose make sure that they adapt themselves
to the user behavior, and at the same time, try to manipulate user
behavior.
Initial Results
We have performed
extensive simulation study [2]
of our proposed mutual
exclusion algorithms. We
came up with three types of
user behaviors: 1)
disciplined, 2) random, 3)
random with CyS helping. In
the figure NoCS corresponds
to the results when there is
no CyS support. Users act on their own. SPRA-1-2-3 represents three
user behaviors. Np is the number of users. AT is acquire time (difference
between time when the user send a request, and the time at which the
user acquires the resource). This experiment models physical system as
a grid of 8X8 locations, and number of resources are constant (3).A user
takes 3 seconds to move from a location to other.
Future Plans
• Immediate goal is to simulate predicate detection and global state
recording algorithms.
Open Problems
• Currently, we investigated CPSs which operate inside a physical system
such as a hospital or a factory. One can study CPSs which are
combination of multiple such CPSs, such as, a series of hospitals.
Middleware to coordinate among multiple CPSs is required to be
studied.
• The mutual exclusion problem we studied required CyS to locate one
resource for a user. In a more complex scenarios, users may need to
reserve multiple resources, and in that case, CyS should be able to
reserve those resources such that the users movements are minimized.
• The simulation we developed is very naïve. Though it adequately
simulates our algorithms, it can be made more robust by providing
functionality, so that one can design the layout of PhyS, place users in
desired locations, change capabilities of users, algorithms on CyS and
so on.
References
[1] A. D. Kshemkalyani. Immediate detection of predicates in pervasive environments.
Journal of Parallel and Distributed Computing, 72(2):219 – 230, 2012.
[2] S. Gujrati and G. Singh, “Mutual exclusion in cyber-physical systems,” in 1st
International Conference on Sensor Networks, February 2012.
[3] S. Chandran and J. Joshi, “Lot-rbac: A location and time-based rbac model,” in Web
Information Systems Engineering WISE 2005, vol. 3806, pp. 361–375.
[4] C. A. Ardagna, M. Cremonini, E. Damiani, S. D. C. di Vimercati, and P. Samarati,
“Supporting location-based conditions in access control policies,” in Proceedings of the
2006 ACM Symposium on Information, computer and communications security. ACM,
2006, pp. 212–222.
[5] V. Rajamani and C. Julien, “Blurring snapshots: Temporal inference of missing and
uncertain data,” in Proceedings of the IEEE International Conference on Pervasive
Computing and Communications, 2010.
[6] Lee, S.; Yoon, D.; Ghosh, A.; , "Intelligent parking lot application using wireless
sensor networks," Collaborative Technologies and Systems, 2008. CTS 2008.
International Symposium on , vol., no., pp.48-57, 19-23 May 2008
Following slides are backup slides.
TDS vs CPS
Fig. 1: CyS superimposed 4.2.
3.1. NCyS
N1
CyS findspath
requests
sends nearest
for to
a
1 moves to WC1.
on a hospital. WC
WC.
WC1.
CyS
Search Path to WC1

Motivational Example :
Mutual Exclusion Request
1.
2.
3. NCyS
1 starts
moves moving
away
generates
patient
from
alert.patient.
to Room3.
CyS
Aler
t

Motivational Example :
Predicate Detection –
Patient moved to other
location without a nurse
being accompanied.
Mutual Exclusion Algo
WC1:Res
Free, WC2: Free, WC3:
1. N
2.
3. CyS
sends
findsWC1
sends atowards
and
request
path
toto
Free Path to4. moves
takes
reserves
WC1
1
6
to NWC1.
.
to search
some
WC1. 1WC.
other location.
WC1

Challenges
Where
is WC? Request
Global State Recording
NS1.N={N1,N2} ICU1.N={} ICU1.P={P1}
R3.WC={WC2} Where are WC1 and N1?

Challenges
Spatial Model
• Logical vs physical areas
• Fine grained vs coarse grained areas
• Contains relation
• Parent child relation
• Reachability edges
• Path between two areas
• Shortest path between two areas
• Distance between two areas
• PAT: a tree structure of the physical system
Event Model

You might also like