ArticlePDF Available

Physical control framework and protocol design for cyber-physical control system


Abstract and Figures

A cyber-physical system is an integration of computation, networking, and physical processes. This article introduces a novel physical control framework to integrate various devices to form the lower-level abstraction network. Two relevant protocols within this framework are proposed for information exchange between different network environments. Furthermore, a formal verification method for the proposed protocols is discussed. The model-checking tool SPIN is used to model and formally verify such protocols. The properties of the protocols are expressed using linear temporal logic to enable model-checking. Implementation results are presented to provide a deeper understanding of the proposed protocols.
Content may be subject to copyright.
Research Article
International Journal of Distributed
Sensor Networks
2017, Vol. 13(7)
ÓThe Author(s) 2017
DOI: 10.1177/1550147717722692
Physical control framework and
protocol design for cyber-physical
control system
Yi Cai
and Deyu Qi
A cyber-physical system is an integration of computation, networking, and physical processes. This article introduces a
novel physical control framework to integrate various devices to form the lower-level abstraction network. Two relevant
protocols within this framework are proposed for information exchange between different network environments.
Furthermore, a formal verification method for the proposed protocols is discussed. The model-checking tool SPIN is
used to model and formally verify such protocols. The properties of the protocols are expressed using linear temporal
logic to enable model-checking. Implementation results are presented to provide a deeper understanding of the pro-
posed protocols.
Cyber-physical system, formal verification, Zigbee, constrained application protocol, Internet of Things
Date received: 23 November 2016; accepted: 26 June 2017
Academic Editor: George Nikolakopoulos
A cyber-physical system (CPS)
connects the physical
world objects and computing systems. However, the
embedded systems in CPSs have limited computation
and storage capabilities. Cloud computing has been
quite successful in integrating several application areas
including on-demand data centers, remote processing
and storage, big data analytics, and resource virtualiza-
tion. Hence, combining cloud computing with CPS is a
new direction for industrial applications.
In cloud
manufacturing environments, it is important to have
monitoring capability to detect any exception cases in
different sites. Wireless sensor networks (WSN) can
help to sense, inspect, and gather the information of
cloud manufacturing environments.
A generic platform is needed to integrate heteroge-
neous monitoring resources and devices into cloud
CPS. In this work, first, we present a novel physical
control framework (PCF) to integrate different net-
works and devices. Next, we propose two protocols for
supporting visiting Zigbee nodes using constrained
application protocol (CoAP). We specify the models of
the two protocols to perform formal verification using
model-checking tool. In the past, we have
already presented our concept along with an initial
In this expanded article, we elaborate
on the concept, add the protocol implementation for
information exchange between Zigbee and CoAP net-
works. Finally, PCF is implemented as a Web-
of-Things (WoT) solution for resource-constrained
School of Computer Science and Engineering, South China University of
Technology, Guangzhou, China
Guangzhou College, South China University of Technology, Guangzhou,
Corresponding author:
Yi Cai, Guangzhou College, South China University of Technology, No.1
XueFu Road, Huadu District, Guangzhou, Guangdong 510800, China.
Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
( which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (
The rest of the article is organized as follows. In the
‘‘Related work’’ section, we summarize the existing
methods of connecting a WSN with TCP/IP networks.
A brief introduction to the verification tool used in our
experiment is given in section ‘‘Formal verification
tool.’’ Next, our proposed solution is presented in sec-
tion ‘‘PCF.’’ Section ‘‘Communication conversion pro-
tocols’ gives a detailed introduction of the two
adaptation protocols named device network access pro-
tocol (DNAP) and device control transmission proto-
col (DCTP), to bridge a Zigbee network and a CoAP
network together in the framework. Simulation and
verification methods of the two protocols are discussed
in section ‘‘Simulation and verification.’’ A security
communication way is proposed in section ‘‘Security.’
Implementation results on the hardware platform are
presented in section ‘‘Implementation.’’ Finally, conclu-
sions are presented in section ‘‘Conclusion and future
Related work
How to connect a WSN to a traditional TCP/IP net-
work is the first challenge for cloud CPS. Oil and gas
companies traditionally use wired-gas sensing systems
because of their reliability and robustness. Due to the
evolution of digital technologies and wireless communi-
cations, a WSN can be an alternate sensing solution for
the oil and gas industry.
Chao et al.
provided an anal-
ysis regarding the ways in which a WSN can be used in
developing a toxic gas monitoring system. Aliyu et al.
provided a Zigbee technology–based gas safety and
monitoring system. However, such solutions isolated
the WSN from the TCP/IP network.
Recent technology trend is to connect a WSN to the
wider Internet over TCP/IP connections. The
6LoWPAN concept originated from the idea that the
Internet protocol could and should be applied even to
the smallest devices.
The 6LoWPAN defines a method
for the IPv6 packets to be carried on top of low-power
and lossy networks (LLNs). The 6LoWPAN uses the
IEEE 802.15.4 standard. Currently, some model sys-
tems are using 6LoWPAN to join WSN and IP net-
works together. The E-SURAKSHAK
is an example
of one such CPS system related to healthcare, and its
end-to-end bidirectional Internet connectivity is pro-
vided by 6LoWPAN.
UDP is a more suitable transport layer protocol for
communication and real-time data transfer in sensor
data acquisition and remote control application scenar-
ios. Hence, most of the Internet of Things (IoT) com-
munication networks mainly use UDP at the transport
layer. At the application layer, there is also a need for a
protocol to integrate real-world sensor devices to the
Web. Modeling sensors’ RESTful resources can offer
services to the world using the HTTP protocol. Dillon
et al.
analyzed the use of web in IoT with a CPS sys-
tem, and provided a novel WoT framework for CPSs.
summarized the implementation challenges
and security problems associated with the conversion
from IoT to WoT. Most WoT implementations focus
on the data exchange at the application layer, and they
also need to take into account the reliance of the IP
layer. Most RESTful approaches are based on the
HTTP protocol, but that would be too complex for
smart devices. The Internet Engineering Task Force
(IETF) established the constrained RESTful environ-
ments (CoRE) working group to realize the REST
architecture in LLNs. This group offers solutions for
the embedded web servers in constrained smart devices.
One cloud application uses CoAP
to communicate
with the web services deployed on a smart device, in a
similar manner as using HTTP. Each resource still cor-
responds to a unique universal resource identifier
(URI). A research survey and analysis related to CoAP
is presented in Villaverde et al.,
and it was concluded
that CoAP is better suited for LLNs compared to other
RESTful protocols as it introduces lower overhead and
minimizes delays while maintaining reliability. The
work of Thombre et al.
evaluates the performance of
the CoAP stack on a device using the Contiki OS and
Cooja simulator in a 5-hop LLN. Cho et al.
a system to allow communication between HTTP and
CoAP networks. Sheng et al.
proposed a framework
including an efficient device management solution for
WSN based on CoAP. The underlying network of this
framework is based on 6LoWPAN. However, none of
the existing works address the problem of integrating a
WSN with CoAP in a Zigbee environment.
The work of Mitsugi et al.
proposed a framework
to implement UPnP over Zigbee network, and to facili-
tate end-to-end HTTP communication between Zigbee
and IP networks. This work used CoAP as the bridge
protocol. The proposed solution required that the sen-
sor MCU associated with the Zigbee transceiver can
handle CoAP messages. This solution approach is not
suitable for Zigbee devices that cannot run CoAP
stack. However, our proposed solution overcomes this
The authors of the White Paper
recommend using
a Zigbee gateway between Zigbee networks and IP net-
works such as the Internet. This solution required that
the Zigbee gateway must allow interactions between IP
applications and the applications on Zigbee devices,
and provided an extensible set of RPC protocols
(SOAP, REST, and GRIP). But this article does not
describe the CoAP binding method for REST-styled
To address the above-mentioned problem, we pro-
pose a new binding method for CoAP. Our solution
can fulfill the features specified in Zigbee Alliance
2International Journal of Distributed Sensor Networks
White Paper,
related to addressing core IP connectiv-
ity, and connecting the Zigbee networks to IP networks
or remote applications. The proposed framework can
scale easily, and there is no need to use CoAP stack on
every Zigbee endpoint device. A new Zigbee device can
join an LLN without any constraints and register itself
by sending requests using the DNAP format. This ful-
fills the requirement of Zigbee Alliance White Paper
in terms of enabling web service for every Zigbee device
based on CoAP.
CoAP provides an Observing method,
servers to send every resource state change to interested
clients in an active manner. The Observe function
offers the possibility for a client to have an up-to-date
representation of the resource without the client having
to constantly poll for information regarding the new
changes. Compared to the HTTP, this feature is an
enhancement for CoAP. For this reason, we design a
timing mode to implement the Observe function in
Zigbee networks.
We try to find a formal specification and model veri-
fication method for protocol used in IoT. The research
work of Aziz
presents a formal model of the MQTT
protocol based on a timed process algebra called TPi.
The authors also present a statistical analysis by manu-
ally limiting the number of copies of input variables.
This approach is an abstract way to simplify the analy-
sis of the MQTT protocol and has weak describing
ability compared to model-checking tools. Model-
checking technology becomes increasingly prominent
in industrial practice. There are some tools for model-
checking, with different capabilities suited to different
kinds of problems. For example, the tools SLAM and
BLAST are only used for performing statistical analysis
of sequential C programs to check safety properties.
The SPIN tool focuses on interaction between pro-
cesses, and SPIN can also check liveness property of
the protocols. In this sense, SLAM and BLAST are
only supporting a subset of the functionalities covered
by SPIN. The KRONOS aims to verify real-time sys-
tems and is dedicated to observation of real-time prop-
erties. KRONOS introduces clock into system states,
which makes the verification of untimed temporal
properties much less efficient than SPIN.
The work of Capella et al.
proposes a reference
model to standardize a variety of WSN monitoring
platforms which are used to gather information about
processing condition of an IoT system. Our PCF can
be used as the implementation of an off-the-shelf com-
ponent for this model. The control gateway can be used
as a sniffer to capture the WSN messages for the inter-
change layer, and the PCF can be an off-the-shelf com-
ponent to obtain information in a standard way for the
information layer.
Formal verification tool
The SPIN tool can be used for the formal verification
of asynchronous process systems or protocols. We
choose it to trace any logical design errors in our pro-
posed protocol. During the validation process, it sup-
ports all correctness requirements expressible using
linear temporal logic (LTL)
The LTL is a modal temporal logic with modalities
referring to time. The LTL is built up from a finite set
of propositional variables A, the logical operators set
f:,^,_,!,$g, and the temporal modal operators
set f½(always), \.(eventually), N(next), U(until),
W(weaklyuntil)g. The symbol denotes ‘‘satisfy.’’
Definition 1. Formally, the set of LTL formulas over A
is inductively defined as follows:
1. True and False both are LTL formulas;
2. If p2A, then pis an LTL formula;
3. :C,C^F,C_F,C!F,C$F,½C,
\.C,NC,CUF, and CWFare LTL formulas
if and only if (iff) Cand Fare LTL formulas.
Definition 2. Let AP be a non-empty set of atomic pro-
positions. A Kripke structure is a four-tuple defined as
M=(S,s0,R,L), where
1. Sis a finite set of states;
2. s02Sis an initial state;
3. RS3Sis a transition relation, for which it
holds that 8s2S,9s02S,and(s,s0)2R;
4. L:S!2AP is a function which labels each state
with the atomic propositions which hold true in
that state.
A finite transition system is a Kripke structure of
M=(S,s0,R,L). An execution of the finite transition
system is an infinite sequence of its states p=s0,
s1,s2::: called path, where 8si2S,(si,si+1)2R,and
Definition 3. Suppose M=(S,s0,R,L) is Kripke
Structure and p=s0,s1,s2::: is a path in M, the rela-
tion pfis defined inductively as follows:
1. ppiff p2L(p0), p2AP;
2. p:Ciff not pC(p6 C);
3. pC_Fiff pCor pF;
4. pC^Fiff pCand pF;
5. pNCiff p1C(p1=s1,s2:::, in the next
time step Cmust be true);
6. pFUCif 9i0,piCand 8k,
Cai and Qi 3
Thus, an LTL formula fholds in sif and only if it
holds for all paths starting at s. If a formula fholds in
s0, it is denoted by Mf.
We use SPIN to verify the protocols’ general proper-
ties of reachability, deadlock freeness, livelock freeness,
termination, and consistency. For this purpose, correct-
ness requirements expressions in LTL are used with
SPIN to trace the model of our data communication
protocols in PCF.
PCF is a framework to integrate different devices and
LLNs together. Figure 1 shows the details of the PCF
model. PCF hides device implementation details from
the higher-level application, and provides logical inter-
faces for physical objects. Based on IP protocol, PCF
offers uniform access interface constituted by physical
control unit (PCU), CPS control gateway layer, and
heterogeneous LLNs. With overlay network technolo-
gies, PCF can be extended or shortened on demand and
make the control crossing physical boundaries between
network devices. All control-related operations are exe-
cuted by the physical control point (PCP).
PCU is an access and control sub-platform for device
control. PCU receives the task descriptions from the
application layer, and manages the device registrations
and cancelations as a generic access network controller.
PCU follows the instructions from the application to
provide synchronous and asynchronous control for
specific devices and assigns them to the corresponding
Device control command conversion module
Device control command conversion module (DCM)
transforms an application layer command to a control
command for the devices. At the application layer,
CPS has no need to consider the underlying differences
among devices and subnets. Application processes use
a unified scripting language to describe the details of a
task and data scheme. DCM transforms the scripting
language into an appropriate hardware platform task
description language. So, the underlying network will
be transparent to the higher-level application in the
CPS control gateway
The CPS control gateway acts as a connector between
the PCU and the underlying heterogeneous networks.
Devices can connect to the CPS control gateway
through Zigbee interfaces. Once the communication
occurs between a sensing domain protocol and a net-
work domain protocol, the gateway needs to execute a
protocol conversion. With the action of PCP, the CPS
gateway could translate the data in the CoAP format
from the side of computer network to another format
on the WSN side.
If the clients on the Internet use HTTP instead of
CoAP, the HTTP-CoAP proxy
can be deployed in
the PCF.
Physical control point
A PCP receives an applications control command in
CoAP format and forms a control message with the
DCTP format. Next, the PCP sends the message to a
destination device. The CPS control gateway works
with the PCP and forwards the message to the final
destination node. If a device needs to send data to an
application, PCP will execute a protocol conversion
again. PCP runs as a CoAP server and responds for the
CoAP client in the IP network.
Communication conversion protocols
We design two protocols to provide an adaption layer
for Zigbee protocols. They work as bridge protocols
and can be used to solve the control for smart devices,
transformation of protocols, and inter-node communi-
cation problems of the PCF. Such devices can be sen-
sors, actuators, coordinators, or gateways.
DNAP is responsible for the access and exit of various
types of equipment from the PCF. The operations of
Figure 1. Physical control framework.
4International Journal of Distributed Sensor Networks
DNAP include device registration, address assignment,
and device disconnection. The PCP automatically cre-
ates a device list to store the access information of
devices at this point. The work-flow of DNAP is speci-
fied below:
1. New devices to join PCF
Whenever a new device boots up, first, it joins a
Zigbee network and receives its 16-bit network
address. This network address is also used as
device ID. The device then sends an access
request to CPS gateway to establish a connec-
tion with a PCP. The PCP then sends a
CONACK message back to the device contain-
ing its ID. Furthermore, the device sends its
detailed resource information to the PCP to do
resource registration.
2. Device to exit the PCF
Initiative exit: If a device requests to quit the
PCF, the PCP disconnects the device after
receiving the request. Next, it releases the sys-
tem resource and reports that the device exit is
Passive exit: When the device is disconnected or
failed, the PCP cannot receive response packets
on time. Next, the PCP issues a disconnect
request and waits for a reply message until a
time-out happens. If a time-out occurs, the PCP
releases the system resource and deletes the
device information from device list. Message
type of the protocol data unit (PDU) is specified
in Table 1.
The PDU format of DNAP is shown in Table 2.
When a device needs to join the PCF, it can use
DNAP to initialize an access request. Once the device
has a successful access to the PCF, a PCP will be
assigned to it. The PCP becomes an agent for the
device and provides multi-tasking support for the
higher-level application layer.
The IEEE 802.15.4 network consisting of tiny inexpen-
sive autonomous devices equipped with sensors is a key
technology for CPS to obtain physical data from real
world. In a WSN, each sensor uses a 64-bit MAC
address which is different from the IP address. In many
Zigbee networks, there is no convergence mechanism
to provide seamless communication between a WSN
and an IP network. For this reason, we propose the
DCTP to coordinate the control and execution.
While traditional Zigbee networks comprise any-
where between 5 and 20 nodes, we can form an LLN of
this size and combine multiple LLNs into a large-scale
WSN as needed. Every LLN is connected by a CPS
control gateway. The gateway plays an important role
of protocol conversion between a Zigbee LLN and an
IP network. By using the DCTP message format, the
control command can be sent to a destination device,
and the sensing data or event notification message
from the device can be sent back to the higher-level
application layer. A CPS gateway connects with the
coordinator in a Zigbee network. PCF uses DHCP
server to dynamically assign IP address to a new CPS
gateway. The application uses this IP address to com-
municate with devices in theLLN.PCPtransformsall
the control commands from DCTP format to CoAP
format, and the corresponding response data from
CoAP format to the DCTP format, respectively. CPS
gateway transmits DCTP message to target device
using its ID. The PDU of DCTP is designed as speci-
fied in Table 3
There are two types of directions for the communi-
cation between the PCP and a WSN. The first direction
is called control direction, and it consists of
Table 1. Message type of DNAP protocol data unit.
PDU type Function
Request PDU CONNECT Initiates a connection request to the gateway
DEVINFO Send information to the gateway device.
DISCONNECT Disconnect the device.
Response PDU CONACK Response for CONNECT PDU, and send device ID back.
ACCEPT Notify device registration is complete.
ACK Device confirm registration have done.
DROPACK Confirm equipment revoked.
Table 2. DNAP protocol data unit format.
Message type PCP ID Device ID Seq. no. Length Data
Cai and Qi 5
communication from PCP to WSN. The other direc-
tion is from WSN to the PCP, called data direction.
1. Control direction
In this case, the PDU is called a control PDU.
The source ID is from the CPS control gateway,
and the destination ID is from the destination
Zigbee device. The value of port domain is used
for correspondence with specific control applica-
tions. Timestamp value is dependent on the
implementation of the sensor system. There are
two time control models. First one is the control-
response mode, and the second one is the timing
mode. Once the application decides to execute
control-response mode, the timestamp value is
set to the time at which a command is sent. In
timing mode, timestamp value is set to zero. The
next domain contains two types:
(a) Specified control: This is represented using
binary format number ‘01’ or ‘10’. This con-
trol format is used to communicate with
specific resources. The control command is
organized into several resource-name/com-
mand pairs and stored in the data domain.
Then the coordinator distributes it to the
destination device. ‘01’ type is for normal
communication and ‘10’ is for single point
(b) Broadcast control:Thisisrepresentedusing
binary format number ‘11’ or ‘00’. This con-
trol message format is used to communicate
with all devices in the WSN. The control
command is available for all WSNs.
2. Data direction
In this case, the PDU is called a data PDU and it
is used to transfer data related to the sensing data
and device status to the PCP. The source ID is
from Zigbee device, and the destination ID is for
the CPS control gateway. The value of port
domain is the port number which is obtained from
the control PDU. Timestamp value is dependent
on the timestamp of the received control PDU. If
the device is in control-response mode, the time-
response data is sent to the CPS gateway. In the
timing mode, the timestamp value is set to zero.
Then, the next domain type chooses data, and it
also has two types:
(a) Response to the specified control PDU: This
is represented using binary format number
‘01’ or ‘10’. The coordinator of a WSN col-
lects the sensing data and device status from
specific devices. The coordinator organizes
this information into several resource-name/
data pairs, stores in the data domain, and
sends this information to the PCP. ‘01’ type
is for single point response and ‘10’ is for sin-
gle point observe response.
(b) Bulk transfer: This is represented using bin-
ary format number ‘11’ or ‘00’, and is used
to transfer original acknowledgement data
collected from a large WSN or the result of
data fusion with compressive sensing theory.
‘11’ type is for multi-point response and ‘00’
is for multi-point observe response.
The PDU type of DCTP is shown in Table 4. PCP
receives a command from the application layer, and
translates the command to a CTRDEV PDU, and
sends it to the coordinator on the gateway. The coordi-
nator periodically transmits the acknowledgement data
back to the PCP using ACKDATA PDU under timing
mode or the sensing data under control-response mode.
Simulation and verification
We use SPIN to simulate and verify the DNAP and
DCTP protocols described in the previous section, and
to trace whether they have any logical design errors.
Analysis and modeling of DNAP
By setting a PCP process and a group of WSN coordi-
nators, we model the DNAP in SPIN to simulate vari-
ous conditions such as device registration, loss of
messages, and retransmission. Simulation result is
shown in the message sequence diagram of Figure 2.
1. Devices register sequence diagram
As shown in Figure 2, five devices try to join
the PCF at random time through a coordinator
Table 3. Protocol data unit of DCTP.
PDU type Source ID Destination ID
Port Timestamp Control/data type Length of data
Table 4. PDU type of DCTP.
PDU type Function
Request PDU CTRDEV Send a group of control instructions to the coordinator.
Response PDU ACKDATA Transfer the sensation data back periodically or responsively.
6International Journal of Distributed Sensor Networks
labeled the TCPoint. Most of the devices can
complete all the registration steps at the begin-
ning. Also, we purposefully introduced features
of loss of messages and retransmission of mes-
sages for some devices, and these devices still
could fix such problems and finish all the regis-
tration steps without any difficulty.
2. Message loss and retransmission analysis
For the case of packet loss at the device side,
the simulation result is shown in Table 5,
whereas the result for the case of packet loss at
the PCP side is shown in Table 6. From the
results, it can be noted that the response for
packet loss errors or retransmissions is handled
correctly under the DNAP protocol.
3. Verification
DNAP verification is done using SPIN. The
results of the verification process are shown in
Figure 3. The three plus (+) symbols in the fig-
ure used to point out that we choose three veri-
fication methods to verify for any invalid end
states, acceptance cycles, and assertion viola-
The tool has checked 138,845 states and found
no errors. It means that there are no deadlocks,
and the verification model for DNAP meets the
safety requirements. By checking the acceptance
cycles, this model did not violate the liveness
In an initial version of DNAP, we did not con-
sider the packet losses. When verification was
executed, we observed a deadlock. After tracing
the error location using the XSPIN message
flows, we found out the reason. Then, retrans-
mission mechanism and timeout process were
added to improve the initial version of DNAP.
The improved version of DNAP got through
the second time verification without any errors
as shown in Figure 3.
Analysis and modeling of DCTP
We model two work patterns of DCTP for different
1. Control-response mode
This mode works in a request-response mode. A
reply associated with the sent control data is
received only if the timestamp is within the
range of the control window. So, the application
and the device process tasks asynchronously.
The timestamp is also used for the heartbeat
detection. So, the PCP can check whether the
sensors are offline. During the data transfer, the
two sides perform the following actions.
(a) The PCP side maintains the latest time-
stamp value that has been sent to WSN,
Figure 2. Simulation results of DNAP.
Figure 3. Verification result of DNAP.
Table 5. Simulation result of packet loss at the device side.
The device side The PCP side
CONNECT drop Status: Cannot get CONACK packet from PCP in time.
DNAP resolution: Re-send it several times if timeout.
Status: Haven’t received CONNECT packet.
DNAP resolution: wait.
DEVINFO drop Status: Cannot get ACCEPT packet from PCP in time.
DNAP resolution: Re-send it several times if timeout.
Status: Haven’t received DEVINFO packet.
DNAP resolution: Wait for some time, give up
this register if timeout.
Cai and Qi 7
and this value is denoted by St. Whenever
the underlying WSN transmits data
through the coordinator, the PCP needs to
check the received timestamp (T) to decide
whether to receive or reject the data.
1. If St W\TSt, PCP receives the data.
2. If TSt W, the received data are obsolete,
data are dropped.
3. If T.St, it is an abnormal data packet and is
In the above logical operations, Wis the maxi-
mum receive window size.
(b) The sensor side maintains the latest time-
stamp value it received from the PCP, and
this value is denoted by NT. Whenever the
sensor receives a control data packet from
the PCP, it updates the value of NT, and
sends out the appropriate control signals
to sensors.
We use LTL to describe the properties and behavior
of DCTP under control-response mode.
Definition 4. The execution of DCTP can be repre-
sented by a Kripke structure of M==(S,s0,R,L),
1. Sis a finite set of protocol states of DCTP.
2. s02Sis an initial state of DCTP.
3. RS3Sis the set of all state transition of
DCTP, for which it holds that 8s2S,9s02S,
and (s,s0)2R,(s,s0) is one state transition.
4. L:S!2AP is a function which labels each state
of the DCTP with the atomic propositions
which hold true in that state.
5. AP =ff1,f2,f3,f4,c1,c2g.
f1: PCU receives a datagram from a device.
f2: The timestamp of the received datagram is
greater than the latest timestamp value that
PCP has sent.
f3: The timestamp of the received datagram is
within the receive window.
f4: The timestamp of the received datagram is
outside of the receive window.
c1: PCP drops the datagram.
c2: PCP receives the datagram.
Proposition 4.1. At any time, if the PCP receives a data-
gram whose timestamp is greater than the latest time-
stamp value that PCP has sent, then PCP drops the
datagram. For this scenario, the LTL formula is given
Proposition 4.2. At any time, if the PCP receives a data-
gram whose timestamp is within the receive window,
and then PCP can receive the datagram. For this sce-
nario, the LTL formula is given by
Proposition 4.3. At any time, if the PCP receives a data-
gram whose timestamp is outside the receive window,
and then PCP must drop the datagram. For this sce-
nario, the LTL formula is given by
An execution of the DCTP is an infinite sequence of
its states p=s0,s1,s2:::. We have to check 8p2M,
whether pholds in P,Qand Hin SPIN
SPIN converts the LTL formulas on the left-hand
side of Figure 4 into three PROMELA never-claims: P,
Q, and H. Then, the SPIN tool performs three model
checks to verify whether properties of P, Q, and H hold
true in the given protocol.
The model-checking operation performs an exhaus-
tive search over all the states reachable by the transit
process. On our verification testing computer, the
complete description of a global system state for
checking every model required 524 bytes of memory
Table 6. Simulation result of packet loss at the PCP side.
The device side The PCP side
CONNECT drop Status: Cannot get CONACK packet from PCP
in time.
DNAP resolution: Re-send CONNECT packet
several times.
Status: Haven’t received DEVINFO packet.
DNAP resolution: Wait for some time, give up this
register if timeout.
ACCEPT drop Status: Cannot get ACCEPT packet from PCP in
DNAP resolution: Wait.
Status: Cannot receive ACK packet in time.
DNAP resolution: It has deal with it at the device
side, so to avoid network congestion DNAP done
nothing at the PCP side.
8International Journal of Distributed Sensor Networks
(per state). The longest depth-first search paths con-
tain 645, 537, and 645 transitions from the root of the
tree for P, Q, and H, respectively. As shown in the
right-hand side of the figure, no errors were found in
these searches.
Figure 5 shows examples of three kinds of message
flow in DCTP. It can be observed from Figure 5(a) that
if the timestamp (line 93, T = 1000, W = 10, St =
1001) is within the receive window, the corresponding
message will be received. As shown in Figure 5(b), if
the timestamp is outside of the receive window (line
133, T = 982, W = 10, St = 1002), the corresponding
message will be dropped. Furthermore, as shown in
Figure 5(c), if the timestamp is abnormal (line 347, T
= 1030, W = 10, St = 1010), the corresponding mes-
sage will be dropped and error handling mechanism
will be triggered.
2. Timing mode
Sometimes, applications need to collect
data continuously from WSN on a long-term
basis. For this purpose, we propose the timing
mode to provide support for the Observe
option of CoAP in the Zigbee LLN. The data
type for the timing mode PDU can be ‘‘01’’ or
‘‘11’’ in the binary format. The binary format
‘‘01’’ is used to send data from the specific sen-
sor that is specified in the control type PDU.
The binary format ‘‘11’’ is used to periodically
transfer the results of multi-sensor data in an
We verify the timing mode of the protocol in
SPIN using all possible states. The verification
result is shown in Figure 6. The result of the
figure indicates that the protocol did not
encounter any deadlocks. In observe way, the
first message sent by client is of control type
must be set to ‘00’ or ‘10’. The notification
response data type can be ‘00’ or ‘10’ in the
subsequent cycles. The binary value ‘10’
implies single point observe, and the value ‘00’
means multi-point observe. The SPIN tool set
up the model-checking for five sensors in an
Figure 4. Verification results under control-response mode.
Figure 5. Simulation result examples of three kinds of message
flow in DCTP: (a) timestamp T:St W\TSt, (b) timestamp
T:TSt W, and (c) timestamp T:T.St.
Figure 6. Verification results under timing mode.
Cai and Qi 9
LLN and performed two types of verification,
and the tool did not find errors. This implies that
no deadlocks were encountered. By checking
acceptance cycles, this model did not violate the
liveness property. The Observe method can be cor-
rectly executed using the DCTP message format.
In order to facilitate transmission of sensitive infor-
mation, PCF permits communication via secure
encryption. When a Zigbee device joins an LLN, PCF
performs an encrypted node authentication. The
coordinator is configured to deny network access to
un-authenticated nodes. Devices and the coordinator
use the pre-shared key (PSK) to establish authentica-
tion and encrypt the communication channel based
on AES-128 in the LLN. DNAP and DCTP messages
are encrypted and encapsulated within the payload of
a Zigbee message. After PCP executing a protocol
transform, the data coming from an LLN can be
transmitted via secure CoAP. To protect CoAP trans-
missions, datagram TLS (DTLS) has been proposed
as the primary security protocol. PCP and client per-
form a handshake to establish a DTLS session. DTLS
message in the form of application data type can be
used to transmit encrypted information between
source and destination for a CoAP application.
Figure 7. Request/response delay compared to normal Zigbee communication.
Figure 8. CoAP observe communication for Zigbee node.
10 International Journal of Distributed Sensor Networks
Our prototype system is built on top of a Zigbee net-
works. Each LLN is formed by five CC2530 nodes. The
comparison results for Request/Response mode are
shown in Figure 7. Four tests were conducted with the
sleep time of Zigbee nodes varying from 1 to 4s. Each
test is performed 50 times. The additional time overhead
compared to normal Zigbee communication was calcu-
framework did not result in a significant overhead to
convert communication from one format to another.
With PCF, the Firefox application with Copper plu-
gin can access sensors using URI ‘‘coap://domain/
resourceName.’’ Applications based on service-oriented
architecture (SOA) can also easily access sensors using
URI under CoAP. Figure 8 shows the CoAP Observe
mode to subscribe data from a sensor. The test dura-
tion was 5 min, and the test statistics confirmed that no
status-reporting messages were lost. From this observa-
tion, it can be concluded that the PCF can implement
the Observe effect on top of Zigbee networks. So, PCF
is a WoT solution for resource-constrained IoT devices.
Conclusion and future works
In this article, we proposed a solution for providing a
universal communication platform for CPS. In order to
integrate multiple protocol devices into framework, we
provided the DNAP to implement a smart device regis-
ter. We also designed the DCTP to facilitate communi-
cation between sensors and the application layer. This
can be viewed as a CoAP solution for upgrading IoT to
WoT network. We also used the SPIN model-checker
tool to perform the simulation and verification for the
proposed protocols DNAP and DCTP.
Possible future works include designing protocols to
facilitate the 3G devices to join the PCP, extending the
PCF to integrate scripting support for the application
logic, and attempting to construct a self-organizing
overlap network that is overlaid on a WoT network.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
The author(s) received no financial support for the research,
authorship, and/or publication of this article.
1. Lee EA. Cyber physical systems: design challenges. In:
Proceedings of the 11th IEEE international symposium on
object and component-oriented real-time distributed com-
puting (ISORC), Orlando, FL, 5–7 May 2008, pp.363–
369. New York: IEEE.
2. Chaaˆ ri R, Ellouze F, Koubaˆ a A, et al. Cyber-physical
systems clouds: a survey. Comput Netw 2016; 108:
3. Bell-Labs. Verifying multi-threaded software with spin, (accessed 25
April 2016).
4. Cai Y and Qi D. Control protocols design for cyber-
physical systems. In: Proceedings of the advanced
information technology, electronic and automation control
conference (IAEAC), Chongqing, China, 19–20 Decem-
ber 2015, pp.668–671. New York: IEEE.
5. Reza M, Talevski A, Talevski S, et al. Applications of
wireless sensor networks in the oil, gas and resources
industries. In: Proceedings of the 24th IEEE international
conference on advanced information networking and appli-
cations, Perth, WA, Australia, 20–23 April 2010, pp.941–
948. New York: IEEE.
6. Chao X, Dargie W and Lin G. Energy model for h2s
monitoring wireless sensor network. In: Proceedings of
the 11th IEEE international conference on computational
science and engineering,Sa
˜o Paulo, Brazil, 16–18 July
2008, pp.402–409. New York: IEEE.
7. Aliyu F, Al-Shaboti M, Garba Y, et al. Hydrogen sulfide
(h2s) gas safety system for oil drilling sites using wire-
less sensor network. Procedia Comput Sci 2015; 63:
8. Mulligan G. The 6LoWPAN architecture. In: Proceed-
ings of the 4th workshop on embedded networked sensors,
Cork, 25–26 June 2007, pp.78–82. New York: ACM.
9. Rao IH, Amir NA, Dagale H, et al. e-surakshak: a
cyber-physical healthcare system with service oriented
architecture. In: Proceedings of the 2012 international
symposium on electronic system design (ISED), Kolkata,
India, 19–22 December 2012, pp.177–182. New York:
10. Dillon TS, Zhuge H, Wu C, et al. Web-of-things frame-
work for cyber–physical systems. Concurr Comp: Pract E
2011; 23(9): 905–923.
11. Raggett D. The web of things: challenges and opportuni-
ties. Comput 2015; 48(5): 26–32.
12. Shelby Z, Hartke K and Bormann C. The constrained
application protocol (CoAP). Proposed Standard, RFC
7252, June 2014,
13. Villaverde BC, Pesch D, Alberola RDP, et al. Con-
strained application protocol for low power embedded
networks: a survey. In: Proceedings of the sixth interna-
tional conference on innovative mobile and internet services
in ubiquitous computing, Palermo, 4–6 July 2012, pp.702–
707. New York: IEEE.
14. Thombre S, Islam RU, Andersson K, et al. Performance
analysis of an IP based protocol stack for WSNs. In:
Proceedings of the IEEE conference on computer com-
munications workshops (INFOCOM WKSHPS),San
Francisco, CA, 10–15 April 2016, pp.360–365. New
York: IEEE.
15. Cho G, Chun S, Jin X, et al. Enhancing CoAP proxy for
semantic composition and multicast communication. In:
Adjunct proceedings of the 2015 ACM international joint
Cai and Qi 11
conference on pervasive and ubiquitous computing and pro-
ceedings of the 2015 ACM international symposium on
wearable computers, Osaka, Japan, 7–11 September 2015,
pp.205–208. New York: ACM.
16. Sheng Z, Wang H, Yin C, et al. Lightweight management
of resource-constrained sensor devices in internet of
things. IEEE Internet Things 2015; 2(5): 402–411.
17. Mitsugi J, Yonemura S, Hada H, et al. Bridging UPnP
and Zigbee with CoAP: protocol and its performance
evaluation. In: Proceedings of the workshop on Internet of
Things and service platforms, Tokyo, Japan, 6–9 Decem-
ber 2011, pp.1–8. New York: ACM.
18. Zigbee Alliance. Understanding Zigbee gateway: how
Zigbee extends an IP network. Zigbee Alliance White
Paper, September 2010.
19. Hartke K. Observing resources in the constrained applica-
tion protocol (CoAP). Proposed Standard, RFC 7641,
September 2015,
20. Aziz B. A formal model and analysis of an iot protocol.
Ad Hoc Netw 2016; 36: 49–57.
21. Capella JV, Campelo JC, Bonastre A, et al. A reference
model for monitoring IoT WSN-based applications. Sen-
sors 2016; 16(11): 1816.
22. Huth M and Ryan M. Predicate logic. In: Huth M and
Ryan M (eds) Logic in computer science: modelling and
reasoning about systems. Cambridge: Cambridge Univer-
sity Press, 2004, pp.93–175.
23. Bormann C, Castellani AP and Shelby Z. CoAP: an
application protocol for billions of tiny internet nodes.
IEEE Internet Comput 2012; 16(2): 62–67.
12 International Journal of Distributed Sensor Networks
Marine Internet of Things (Marine IoT) attracts growing interest in sensing and exploiting the ocean, it is necessary to build Marine IoT system for observing and processing massive marine environmental data. There exist several problems in building Marine IoT system, such as massive heterogeneous data, underwater limited transmission bandwidth, strong heterogeneity and unbalanced computing-resource load of data-sensing devices. In this paper, the RESTful-Edge Computing (REC)-based Marine IoT model is proposed to achieve efficient integration and procession of massive marine observing data. The REC-MPSO-based computing resource cooperative migration method is then provided for scheduling and migrating computing resource of Marine IoT. (1) The REC model would store and analyze massive data, transmit only valid observing data to the cloud server, to reduce amount of data transmission. (2) By unified scheduling and management of computing resource load, the edge gateway can realize migration of resource loads between different edge sensing devices, thereby maintaining load balance on these devices. (3) The optimal computing resource (task) for cooperative migration could be selected by Multi-group PSO (MPSO) update iteration. It would realize the dynamically adaptive matching between computing tasks and edge sensing devices, and meet the requirements of minimizing energy consumption and delay of computing task migration in Marine IoT. Simulation results verify that the improvement of proposed REC-MPSO method is 10%, 14% and 22% over three pre-existing protocols in terms of time delay, energy consumption of REC-MPSO is 17%, 24%, and 27% lower than three pre-existing protocols for computing resources migration, respectively.
The research proposes a metric referred to as Smart Manufacturing Performance Measures (SMPMs), which are mapped with manufacturing decisions (structural and infrastructural) and manufacturing outputs (cost, quality, delivery, flexibility, and innovativeness). The mapping between manufacturing decisions, manufacturing outputs, and SMPMs, facilitates the organization in choosing SMPMs to deliver desired output as represented by the competitive priority. The approach provides a direction for the practitioners to implement and operate smart manufacturing systems efficiently and effectively by devising a performance measurement system. A total of 22 performance measures are identified through an in-depth literature review. These are further classified into hard and soft measures, which allow decision-makers to measure SMPMs and focus on a specific performance measure. Further, SMPMs is linked to 13 enablers, which are the driving technologies for implementing smart manufacturing systems. Smart manufacturing is evolving, and its implementation is still in the nascent phase. The proposed approach plays a vital role in redefining the manufacturing strategy while deploying smart manufacturing systems. The research contribution is a step-in assessment of smart manufacturing to understand the ‘what’ and ‘how’ of technological changes are transforming the operations and driving outputs.
Conference Paper
Data analytics using real-time data has brought many challenges, particularly in respect to security. From data generation to storage, each step in the life cycle has its own challenges. This paper identifies those challenges. There are several security options that can utilized but for the purpose of this paper we have limited ourselves to the two most common mainstream techniques already used for protecting big data: encryption and access control. The goal is to ensure the confidentiality, integrity and availability (CIA) for phases of the data life cycle, from generation to storage. This research offers a novel approach to the data security and of real-time data transfer and analytics. This paper presents metrics for each of the identified security challenges and aims to validate a real-time data transfer and analytics environment from the security perspective of encryption and access control.
Full-text available
O presente trabalho destina-se a esclarecer o conceito de Indústria 4.0 para sistemas de controle industrial de Engenharia Química, não abandonando as técnicas consagradas e colocando-as no contexto ciber-físico através da integração entre sistema de informação, modelo virtual e o processo físico. Após a elaboração da arquitetura e programação dentro de cada módulo de sistema de controle, incluindo um observador virtual, é realizada a aplicação de intertravamento no controle de um reator de batelada, mostrando as funcionalidades de automação ciberfísica quando comparada à complexidade de hardwares em sistemas tradicionais.
Full-text available
The Internet of Things (IoT) is, at this moment, one of the most promising technologies that has arisen for decades. Wireless Sensor Networks (WSNs) are one of the main pillars for many IoT applications, insofar as they require to obtain context-awareness information. The bibliography shows many difficulties in their real implementation that have prevented its massive deployment. Additionally, in IoT environments where data producers and data consumers are not directly related, compatibility and certification issues become fundamental. Both problems would profit from accurate knowledge of the internal behavior of WSNs that must be obtained by the utilization of appropriate tools. There are many ad-hoc proposals with no common structure or methodology, and intended to monitor a particular WSN. To overcome this problem, this paper proposes a structured three-layer reference model for WSN Monitoring Platforms (WSN-MP), which offers a standard environment for the design of new monitoring platforms to debug, verify and certify a WSN’s behavior and performance, and applicable to every WSN. This model also allows the comparative analysis of the current proposals for monitoring the operation of WSNs. Following this methodology, it is possible to achieve a standardization of WSN-MP, promoting new research areas in order to solve the problems of each layer.
Full-text available
Cyber-Physical Systems (CPSs) represent systems where computations are tightly coupled with the physical world, meaning that physical data is the core component that drives computation. Industrial automation systems, wireless sensor networks, mobile robots and vehicular networks are just a sample of cyber-physical systems. Typically, CPSs have limited computation and storage capabilities due to their tiny size and being embedded into larger systems. With the emergence of cloud computing and the Internet-of-Things (IoT), there are several new opportunities for these CPSs to extend their capabilities by taking advantage of the cloud resources in different ways. In this survey paper, we present an overview of research efforts on the integration of cyber-physical systems with cloud computing and categorize them into three areas: (1) remote brain, (2) big data manipulation, (3) and virtualization. In particular, we focus on three major CPSs namely mobile robots, wireless sensor networks and vehicular networks.
Conference Paper
Full-text available
Wireless sensor networks (WSNs) are the key enablers of the internet of things (IoT) paradigm. Traditionally, sensor network research has been to be unlike the internet, motivated by power and device constraints. The IETF 6LoWPAN draft standard changes this, defining how IPv6 packets can be efficiently transmitted over IEEE 802.15.4 radio links. Due to this 6LoWPAN technology, low power, low cost microcontrollers can be connected to the internet forming what is known as the Wireless embedded internet. Another IETF recommendation, CoAP allows these devices to communicate interactively over the internet. The integration of such tiny, ubiquitous electronic devices to the internet enables interesting real-time applications. We evaluate the performance of a stack consisting of CoAP and 6LoWPAN over the IEEE 802.15.4 radio link using the Contiki OS and Cooja simulator, along with the CoAP framework Californium (Cf).
Conference Paper
Full-text available
Due to the resource constrained nature of things on the Internet of Things (IoT), a mechanism of managing things and processing their data needs to be low-cost in terms of power consumption and network traffic. The Constrained Application Protocol (CoAP) is suitable for the IoT and supports multicast communication with things. However, multicast may lead to a traffic overload when the corresponding multiple responses travel through a CoAP-proxy concurrently. In this paper, we propose a novel approach to constructing a CoAP-proxy, which supports the efficient multicasting and grouping things based on the semantic description of a request. In addition, the proposed proxy merges multiple responses into an aggregated one, which is returned to a client. To demonstrate the feasibility, we implemented a prototype of the proposed CoAP-proxy and evaluated the performance of our approach in terms of the response time and network traffic.
Full-text available
Toxic and Combustible gases are one of the most notorious killers of Oil and Gas Industry (OGI) workers. Hardly can one find an accident in which they are not involved, as such governments around the world enforce the use of gas sensors in all works that are related to these gases. Until recently, wired gas sensing systems are used. Although, they are rugged and reliable, they are expensive to install and maintain. Wireless gas leakage detection systems are flexible, easy to install and maintain. However, gas sensors tend to consume lots of energy, while those that consume less energy are not accurate. In this paper, an experimental research is carried out to investigate the energy performance of a Zigbee Wireless Sensor Network (WSN) made up of wireless sensor nodes and industrial-grade gas sensors to create a Wireless Gas Safety and Monitoring System (WG-SMS). The system uses careful combination of energy harvesting, sensor node role assignment and adaptive sleep cycle to achieve maximum attainable lifetime by saving 60% more energy.
Full-text available
We present a formal model of the MQ Telemetry Transport version 3.1 protocol based on a timed message-passing process algebra. We explain the modelling choices that we made, including pointing out ambiguities in the original pro- tocol specification, and we carry out a static analysis of the formal protocol model, which is based on an approximation of a name-substitution semantics for algebra. The analysis reveals that the protocol behaves correctly as speci- fied against the first two quality of service modes of operation providing at most once and at least once delivery semantics to the subscribers. However, we find that the third and highest quality of service semantics is prone to error and at best ambiguous in certain aspects of its specification. Finally, we suggest an enhancement of this level of QoS for the protocol.
Conference Paper
Cyber-Physical System is a relatively new thing with continuous development nowadays. But there is a lack of universal underlying support platform to integrate the underlying devices and screen the technology details. We introduce a Physical Control Framework to integrating real-world devices to form the lower-level abstraction network with complexity and heterogeneity. Then we propose two relevant protocols within this framework to satisfy the need of controlling over remote devices and exchanging information from sensors. Further, we use the model checking tool SPIN to model and to verify such protocols. Some requirements for the protocols are expressed using linear temporal logic to enable requirement analysis and early detection of inconsistencies.
As billions of devices connect to the Internet, a new Web of Things is emerging, with virtual representations of physical or abstract realities increasingly accessible via Web technologies. Achieving a new phase of exponential growth, comparable to the earliest days of the Web, will require open markets, open standards, and the vision to imagine the potential for this expanding WoT.
Conference Paper
In this paper we present the design of "e-SURAKSHAK," a novel cyber-physical health care management system of Wireless Embedded Internet Devices (WEIDs) that sense vital health parameters. The system is capable of sensing body temperature, heart rate, oxygen saturation level and also allows noninvasive blood pressure (NIBP) measurement. End to end internet connectivity is provided by using 6LoWPAN based wireless network that uses the 802.15.4 radio. A service oriented architecture (SOA) [1] is implemented to extract meaningful information and present it in an easy-to-understand form to the end-user instead of raw data made available by sensors. A central electronic database and health care management software are developed. Vital health parameters are measured and stored periodically in the database. Further, support for real-time measurement of health parameters is provided through a web based GUI. The system has been implemented completely and demonstrated with multiple users and multiple WEIDs.