SeSAMe: Software Defined Smart Home Alert

Management System for Smart Communities

Rohit Abhishek1 , Shuai Zhao1 , David Tipper2 , Deep Medhi1
1
University of Missouri-Kansas City, 2 University of Pittsburgh
{rabhishek, shuai.zhao, dmedhi}@umkc.edu, [email protected]

Abstract—Future Smart Cities require new ways to manage to communicate important public safety information/alerts to
services that benefit the end users. An important issue is how to each other, such as fire in any home or an amber alert.
connect homes in a community and create an alert management Whenever an incident like a fire takes place in a home,
system in Smart Cities with coordination among different entities.
In this paper, we present an architectural vision of a software a fire hazard severity zone can be formed around the place
defined home alert management system for Smart Cities. This of the fire. If any home lies within the danger zone, it has
alert management system would make the residents aware of a high probability of catching fire. During these situations, a
any incidents in the neighborhood such as fire. In this work, major concern is to inform the people, who live in these danger
we use the features of software defined networking to design a zones, about the fire. Generally, to do this, the authorities send
manageable and flexible smart home for a smart community to
provide services such as smart alarm systems. out alerts to the public in that zone and broadcast it on the
news as well. However, this might not reach all the people
I. I NTRODUCTION who need to be informed.
Providing smart services in Smart Cities has become an In this work, we propose SeSAMe as an architectural
emerging interest in the past few years. The goal is to improve vision for software defined smart community home alarm
and simplify the life of citizens living in a city by integrating management based on software defined networks (SDN). We
information and communication technology (ICT) and Internet present the protocol messages and system components for
of Things (IoT) solutions to create better service facilities for the operation of SeSAMe. With our approach, should any
the citizens [6]. There are six major components in a smart city alert/event such as a fire occur, an automated notification is
as shown in Fig. 1. In this paper, we focus on smart homes—a sent to all the homes in the neighborhood and to the fire
major component of smart living in smart cities. department and the police department about the fire. At the
A smart home, in its core, can be defined as an automated same time, alerts can also be forwarded to the police and the
home where all devices and appliances are networked together; fire departments.
they coordinate to make intelligent decisions and can be In future smart homes, we anticipate a number of sensors
controlled remotely by the owner of the home. One of the for monitoring a variety of information about the homes. We
important features in home automation is that the home owner believe the use of SDN would give the flexibility of adjusting
decides the reaction of the device. Initially, smart homes were to old and newly added sensors and traffic that arises from
thought to be to controlled environmental systems, but recent them. An advantage of using SDN is that the configuration of
development in technology has enabled it to cover almost any highly complex sensor devices can be made easy through the
electrical device within the home [13]. That is, smart homes centralized SDN controller.
aim to improve the lives of the residents by (1) creating The rest of the paper is organized as follows. Section II
secure homes, (2) saving energy, and (3) improving home presents the related work followed by a brief overview of SDN
accessibility in a convenient and flexible way. A smart com- in section III. Section IV presents our proposed architecture,
munity can be viewed as a virtual environment where smart section V describes the system initialization, and section VI
homes are networked together in a local geographic region to shows our experimental result followed by conclusion and
continuously monitor various aspects of the community and future work in section VII.
provide feedback to improve the safety, security, quality and
emergency response abilities of the community [12]. II. R ELATED W ORK
In a smart home, different electronic devices are networked Li et al. [12] introduces the smart community as a new
together through a Home Area Network (HAN). For example, application of the Internet of Things. In this work, the smart
motion sensors communicate with light sensors and thermal community architecture has been defined and how to realize
sensors to switch lights on/off and adjust the temperature in the secure and robust networking among individual homes has
home. The smart homes can study the resident’s daily activities been described. Xu et al. [14] proposes the software defined
pattern and accordingly adapt to it. smart home (SDSH). The authors use the core idea of SDN
The overall goal of our work is to enable features for a to design an SDSH and list the advantages of SDN such as
smart community where smart homes are able to communicate centralization, optimization, and virtualization in designing the
with each other and exchange information. This can be used smart home. The SDSH discussed in the work focuses on

978-1-5386-0728-2/17/$31.00 c 2017 IEEE

Authorized licensed use limited to: SRM Institute of Science and Technology Kattankulathur. Downloaded on February 13,2025 at 07:33:42 UTC from IEEE Xplore. Restrictions apply.
 global network view via dynamic, automated SDN programs
[5], [9]. The SDN controller, as the control entity of SDN
architecture, manages the network flows to the data plane
that lies on the southbound protocols (ex. OpenFlow) and to
the applications that lie on the northbound interface by using
different API calls [7].
By providing great flexibility of managing network applica-
tion flows, SDN can manage all the network services via the
control plane to allow the dynamic response to network needs
[1], which is a key factor in the prospect of Smart Cities. For
example, when there are many nodes in the network sending
Fig. 1. Smart City [8]
messages simultaneously, our proposed SeSAMe home man-
agement system can control the network based on a global
network view and assign network resources accordingly.
building a feasible system for easy operation and open APIs
IV. S E SAM E : A RCHITECTURE
to connect with third-party services. The work proposed in
this paper deals with managing the network inside the home, The architectural vision of SeSAMe is shown in Fig. 3.
whereas in our work, we propose to use SDN for the network It shows homes that are located in a neighborhood; for each
in a smart community connecting the homes together for home neighborhood, a fog based/localized SDN controller may be
alert management. assigned.
In [10], the authors designed a bandwidth allocation frame- Each home has a home gateway that monitors the reading
work for an SDN based smart home. The bandwidth allocation from the sensors in the home. The home gateway is connected
framework proposed in this work is based on SDN architecture to a centralized SDN controller. The overall network in a
and is able to manage IoT devices for each smart home by neighborhood has a tree structure that is common for con-
designating an ISP to optimize the bandwidth allocation on necting residences to the rest of the system. When any event
both internal home traffic and external Internet traffic. In our such as a fire is detected, the fire sensor sends the reading to
work, we take advantage of the flexibilities of using SDN the home gateway. The home gateway would then try to notify
described in [10] to design a more manageable and flexible the different homes in the neighborhood as well as the police
smart home alert management system for sensors in a smart and fire departments about the occurrence of the event by
home. communicating through the SDN controller. It installs flows on
the route so that the home where the fire has occurred is able
III. SDN:OVERVIEW to notify the other entities. Fig. 4 and 5 depict a fire scenario.
Software-Defined Networking (SDN) provides a dynamic, In Fig. 4, when the sensor detects the fire in one of the homes,
flexible, and controllable platform for making it an important it notifies the home gateway, which in turn attempts to alert
network architecture for the dynamic nature of today’s network different homes in the neighborhood, the fire department, and
applications. In our approach, the underlying network function the police department about the incident. Since it may not
for our smart home management system (SeSAMe) is provided know the route to other homes and the fire and the police
using SDN. SDN is a network architecture where the data deparments, it sends a notification to the controller about the
plane and the control plane are decoupled from each other, and alert, the type of alert (e.g., fire, acts of nature, or theft),
the data plane is managed remotely by the control plane. The its location, and asks it for the route to reach other homes.
four main features of SDN are [9] as follows: (1) control plane
and the data plane are separated, (2) the control plane can be
centralized, (3) the control plane can be programmed, (4) the
application programming interfaces (APIs) are standardized.
Fig. 2 depicts an overview of the SDN architecture design
that shows three layers: the data plane, control plane, and
application plane (management plane). The decoupling of the
data and the control plane makes the network administration
becomes flexible and manageable, as well as significantly
lowers the cost of the physical data forwarding hardware.
The centralization of the control plane has brought in several
significant advantages. As compared with low-level device-
specific configurations, the centralization of the control plane
makes the overall network architecture less error prone to
modify network policies through high-level languages and
software components [11]. It allows the control plane to get a Fig. 2. The SDN Framework

Authorized licensed use limited to: SRM Institute of Science and Technology Kattankulathur. Downloaded on February 13,2025 at 07:33:42 UTC from IEEE Xplore. Restrictions apply.
 Fig. 7. Connection between home and controller

Fig. 3. SeSAMe: Connected Smart Home Architecture

The controller then alerts the police and fire departments and
informs the home about the route to other homes and installs
flows on them as shown in Fig. 5. Once the flows are installed,
the fire affected home sends alerts to the different homes that
include the alert id and its location. Fig. 8. Trigger Event

Fig. 6 shows the high level architecture of a smart home

in SeSAMe. It can be categorized into two categories: home
gateway and sensors. The sensors include different sensors
that are part of the home, e.g., fire sensor, temperature sensor,
light sensor, motion sensor, and so on. All sensors send their
data to the home gateway. The controller creates a database
of the readings from various sensors. As shown in the figure,
there are three sensors in the home and the controller creates a
database for each of the sensors. The home gateway consists of
Fig. 4. Fire notification from Fig. 5. Notification from the fire a database where the reading from different sensors is stored,
home to controller effected home to different homes
at least temporarily.
The management layer is the core of a smart home. It
For system resilience, each sensor in a house is also associ-
continues monitoring the data coming from different sensors.
ated with a secondary home gateway from another home in the
Based on the data collected from the sensors, it creates a
neighborhood. This way, if a sensor’s primary home gateway
triggered event that is sent to the controller along with the
is non-responsive (e.g., the fire disabling the home gateway as
type of data and the reading (or a notification that the fire has
well), then the secondary home gateway can still communicate
been detected).
about the distress message generated by the sensor.
A. Message Types
Fig. 7 depicts the connection between a smart home and
the controller. The smart home continue to send keep alive
messages to the controller at regular intervals to ensure that
the link between the home and the controller is working and
the home is connected to the controller. At the same time,
the controller periodically sends a PingPong message to the
homes to ensure they are connected to the alert system. Keep
alive messages sent by the home gateway are a type of a one
way communication, while PingPong messages are a type of
a two way communication.
In SeSAMe, we define four different types of messages as
shown in Table I.
Fig. 6. Home Architecture • Update messages are sent from the home to the con-
trollers to make it aware of its reachability and any

Authorized licensed use limited to: SRM Institute of Science and Technology Kattankulathur. Downloaded on February 13,2025 at 07:33:42 UTC from IEEE Xplore. Restrictions apply.
 TABLE I kind of alert it is. The Value field represents the reading
M ESSAGE TYPES of the alert device.
Message Type Sender-receiver Description
Update Home-to-Controller Sends updates to
the controller
Trigger Home-to-Controller Event triggered
message to notify
the controller of
any alert
Announce Home-to-Homes Notify the homes
of any alert
Keepalive Home-to-Controller Notify the Fig. 10. Trigger message fields
controller that the
home is connected
• An Announce message is generated by the home where
the event occurred and will be sent to all other homes in
changes made to any device at home. There are two types the neighborhood. When an event takes place in a home, it
of timers used here. One is the update timer and the other notifies the controller. The controller in turn installs flows
one is the holddown timer. The holddown timer value is on the route connecting the affected home to all other
several times the value of the periodic update timer. If homes in the neighborhood. Once the flows are installed,
the updates sent by the homes to the controller are lost, the affected home announces the message to the other
the controller would not assume that the home is down. homes. The announce message fields are shown in Fig.
Instead, it waits until the holddown timer expires. When 11. An alert ID again represents the type of alert. The
no updates are sent by the home to the controller (when GPS ID of the alert home represents the location of the
the holddown timer exipres) it assumes that the devices home where the event occurs.
at this home are not accessible.

Fig. 11. Announce message fields

• Keepalive messages are sent by the homes to the cen-

Fig. 9. Update message fields
tralized controller to check that the connectivity between
them is working or to prevent the link from being broken
The different fields of update messages are shown in Fig. [3]. It is a one way communication. If the controller does
9. We assume that all sensors and the home gateways not receive the keepalive message from the home after a
are equipped with an Indoor Positioning System (IPS) predefined interval, it would mean that either the home or
[2] for an accurate GPS ID of the locations. The GPS the link between the controller and the home is down. It
ID of the home gateway represents the GPS location of is useful in scenarios where a link is down and the traffic
the home. Updates are sent every fixed regular interval. has to be rerouted via another path. The message fields
Active devices is an optional field that represents all the are shown in Fig. 12.
devices that are working properly.
• A Trigger message is generated by the homes to notify
the controller of any alerts. The information in the trigger
message includes the alert type. Trigger events here
denote the conditions when the home gateway sends alerts
to the controller. Fig. 8 shows the triggered events. In
the management layer, the threshold levels for different Fig. 12. Keep-Alive message fields
readings are already defined. Thus, when the management
layer finds any of the readings going above the threshold, The home gateway communicates with the central controller
it creates a triggered event. This triggered event is used to using two protocol PingPong protocols. The PingPong proto-
notify the controller about the event (for example, fire). col is initiated by the controller by sending the first message.
Fig. 10 shows the fields included in the trigger message. After receiving the message, the client, which is the home
Alert ID represents the ID of the sensor reading that is gateway in this case, responds to the server. The PingPong
sent to the controller. Different sensor alerts will have protocol makes sure the home gateway is working and is
a unique alert ID code, which would be attached with connected to the controller. If the controller does not receive
the trigger event packet and sent to the controller. For any response from the home gateway, it first sends a pull
example, the fire alert may have an alert code “101 Fire”. request. If the home does not respond to the request, it assumes
By checking the alert code, the controller knows what that the home gateway is down.

Authorized licensed use limited to: SRM Institute of Science and Technology Kattankulathur. Downloaded on February 13,2025 at 07:33:42 UTC from IEEE Xplore. Restrictions apply.
 The controller keeps listening for the triggered events sent • Normal condition: When the primary home gateway is
by the home gateways. Also, after repeat intervals, it keeps up, the centralized SDN controller receives information
using the PingPong protocol. If any home sends a trigger event from the primary as well as the secondary home gateway.
to the controller, the controller then forwards it the addresses In this case, the information received from the secondary
of the homes where the alert should be broadcasted and also controller is ignored.
installs flows on the route. • Home gateway failure/ Link failure: When the primary
home gateway loses connection to the centralized SDN
B. Sensor Data controller, which can either happen because of the home
Different sensors in the home will send data to the home gateway failure or due to the failure of the link between
gateway and the secondary home gateway assigned to each the home and the centralized controller, the secondary
sensor. Each of the sensor datum will consist of 6 different controller takes the role of the primary controller. So
parameters: timestamp, location, home address, phenomena, the centralized controller would receive information from
value, and unit. The timestamp denotes the time when the the secondary controller as a backup way to update the
reading of the sensor was taken; the location denotes the IPS sensors.
location; the home address would be the address of the primary
gateway to which the sensor is connected; the phenomena A. Issues
represents the kind of reading that is being measured; the value SeSAMe could be an opt-in service. In this case, the home
is the reading of the sensor and the unit is the unit associated owners, by opting-in, agree to share their home locations
with the sensor. and sensor information. It could also include registering the
user’s phone number/numbers that can be used to send alert
C. Policy Rules
messages. While as a service, there are benefits of SeSaMe, it
The central controller also maintains various policy rules. also could present privacy risks for the owner. If the location of
For example, certain types of information available from the the user’s house is shared with the controller and is maintained
sensors at a home may only be sent to the police and the by a third party provider, it raises privacy risks. Services for
fire department, but not to the others in the neighborhood. which the location/information of the user are shared, require
Similarly, if a home does not wish to be a secondary home the use of a secure communication and privacy preservation.
gateway for sensors in the neighborhood, this can be specified If the server becomes a point of attack, the user’s information
through a policy rule. could be compromised.
VI. S IMULATION S ETUP AND P RELIMINARY R ESULTS
V. S YSTEM I NITIALIZATION AND O PERATION The SeSAMe testbed was setup on GENI [4]. The exper-
The SeSAMe setup includes setting up the home gateway, imental topology is shown in Fig. 16. There are 12 homes
and then the sensors are installed. Once the sensors are as shown in the figure. We tested the time to notify different
installed, it sends out a ‘discover’ message to the nearby hosts in the topology. The capacity of each link was set to
gateways. The gateways, after receiving the ‘discover’ mes- 100 Mbps. We tested the normal scenario as well the home
sage, send an ‘offer’ message in reply to it, which includes gateway failure scenario where home 1 loses connection to
the location of the gateways. As there can be more than one the controller and home 2 acts as the secondary controller.
gateway in the neighborhood sending the ‘offer’ message to The result for both the scenarios are the same as both homes
the sensor, it uses the election protocol to select the primary are in the same cluster.
and the secondary gateway. It selects the gateway that is As the message size increased and the number of sensors
closest as the primary gateway (which should be the home sending the message increased, the time to receive the message
where the sensor is installed) and the secondary gateway is the increases. We measured the time it took to send a message
second nearest gateway. After the primary and the secondary from home 1 to home 3 that was on the same node cluster
gateways are selected, the sensors send a ‘request’ message and to home 4 and home 7 that were on different node
to the selected gateways, indicating the gateways to add them clusters. Table II shows our simulation results based on each
as an authorized device. The home gateways send an ‘ack’ case being independently replicated 10 times and ± represents
message reply to the sensor indicating that it has been added the 90% confidence interval. One-to-One means the message
as an authorized device and the gateway is ready to receive was sent to just one host. For example, h1-h3 means when
the data from the sensors. home 1 is sending the alert just to home 3. The number of
As a result of the primary and the secondary home gateways, sensors represents the total number of sensors that are sending
at any particular time the data transmitted by the sensor are messages at a time (with each sensor sending one message).
received by these two gateways. Therefore, the controller One-to-Three signifies the time it takes for the alert to reach
receives information about the sensor status of homes from home 3, home 4, and home 7 when home 1 sends the alerts
the primary as well the secondary controller at any instance to three hosts at the same time. The time for One-to-Three is
of time. A few scenarios shown below detail the use of the greater (as compared to One-to-One) because the alert is sent
primary and secondary controllers: to the three hosts at the same time, so the link gets congested

Authorized licensed use limited to: SRM Institute of Science and Technology Kattankulathur. Downloaded on February 13,2025 at 07:33:42 UTC from IEEE Xplore. Restrictions apply.
 Fig. 13. h1-h3 Fig. 14. h1-h4 Fig. 15. h1-h7

and fire departments. The preliminary results show that using

the SDN approach alerts can be communicated very quickly.
Moreover, they give the flexibility of programmable control
functions, lower operating costs, and centralized management,
to name a few.
In the future, we plan to implement a system of distributed
controllers in place of a single controller when a large neigh-
Fig. 16. Experimental Topology borhood area is involved. Here, all controllers for different
neighborhoods would be connected to each other. Thus, each
as the capacity of the link is limited. The graphs shown in Fig. controller would choose another controller that would act
13, 14, and 15 show the time to receive the message when the as its backup controller. A backup controller will be used
message of 1 MB is sent. For each of the graphs shown, we only when the controller reaches its processing threshold, i.e.,
compared the time to send the message for one-to-one Vs. one- it is no longer able to process any requests. During such
to-three for 1,10, and 100 sensors. We can see that the time scenarios, it will forward all the requests to its designated
to send the message was not too significant (keeping in mind backup controller.
that the capacity of the link was 100Mbps) as the message ACKNOWLEDGEMENT
size increased and the distance to which the message being
This work is partially supported by the National Science
sent also increased.
Foundation Grant # 1526299.
TABLE II
R ESULTS
R EFERENCES
[1] http://www.quotecolo.com/smart-cities-how-sdn-and-nfv-are-changing-
One-to-One (ms) the-way-we-live/.
No of Sensors
[2] https://en.wikipedia.org/wiki/Indoor positioning system.
100 bytes 10000 bytes 1 MB
h1-h3 1 8.467 ±0.65 8.52 ±0.25 84.56 ±0.57
[3] http:https://en.wikipedia.org/wiki/Keepalive.
10 9.83 ±1.10 12.65 ±1.87 626.08 ±57.61 [4] https://www.geni.net.
100 112.86 ±50.71 93.79 ±43.50 6376.47 ±108.16 [5] Software-defined networking (SDN) definition.
h1-h4 1 9.70 ±0.31 9.59 ±0.30 88.90 ±1.24 [6] https://en.wikipedia.org/wiki/Smart city.
10 11.16 ±1.12 13.88 ±5.30 647.58 ±60.66 [7] https://www.sdxcentral.com/resources/sdn/sdn-controllers/.
100 150.10 ±61.41 110.41 ±59.17 6598.44 ±123.40 [8] http://www.smartbrantford.ca/TheSixComponents.aspx.
h1-h7 1 10.80 ±0.17 10.72 ±0.22 96.92 ±4.62 [9] R. Jain and S. Paul. Network virtualization and software defined
10 13.12 ±1.75 16.80 ±0.68 658.74 ±54.68 networking for cloud computing: a survey. IEEE Communications
100 111.18 ±51.46 119.88 ±63.79 6703.80 ±124.43 Magazine, 51(11):24–31, 2013.
One-to-Three (ms) [10] H.-C. Jang, C.-W. Huang, and F.-K. Yeh. Design a bandwidth allocation
100 bytes 10000 bytes 1 MB framework for sdn based smart home. In Information Technology,
h1-h3 1 17.19 ±2.36 19.96 ±2.42 218.75 ±16.56 Electronics and Mobile Communication Conference (IEMCON), 2016
10 117.70 ±10.81 124.19 ±6.00 1632.31 ±84.42
IEEE 7th Annual, pages 1–6. IEEE, 2016.
100 437.66 ±46.40 513.55 ±79.80 19494.4 ±640.81
h1-h4 1 24.70 ±2.59 25.49 ±2.97 229.46 ±14.53 [11] D. Kreutz, F. M. Ramos, P. E. Verissimo, C. E. Rothenberg, S. Azodol-
10 131.11 ±3.60 133.02 ±8.83 1745.22 ±178.19 molky, and S. Uhlig. Software-defined networking: A comprehensive
100 458.76 ±42.14 531.38 ±44.75 19815.8 ±627.82 survey. Proceedings of the IEEE, 103(1):14–76, 2015.
h1-h7 1 27.99 ±3.86 29.54 ±4.72 268.98 ±37.33 [12] X. Li, R. Lu, X. Liang, X. Shen, J. Chen, and X. Lin. Smart community:
10 139.437 ±2.94 133.66 ±7.26 1784.10 ±188.14 an internet of things application. IEEE Communications Magazine,
100 476.729 ±33.73 543.911 ±41.22 22604.9 ±273.35 49(11), 2011.
[13] V. Ricquebourg, D. Menga, D. Durand, B. Marhic, L. Delahoche, and
C. Loge. The smart home concept: our immediate future. In 2006 1st
VII. C ONCLUSION AND F UTURE WORK IEEE international conference on e-learning in industrial electronics,
pages 23–28. IEEE, 2006.
In this paper, we propose SeSAMe, a software defined smart [14] K. Xu, X. Wang, W. Wei, H. Song, and B. Mao. Toward software
home alert system, based on SDN, which is an automated sys- defined smart home. IEEE Communications Magazine, 54(5):116–122,
2016.
tem to send notification alerts to other homes, and the police

Authorized licensed use limited to: SRM Institute of Science and Technology Kattankulathur. Downloaded on February 13,2025 at 07:33:42 UTC from IEEE Xplore. Restrictions apply.