Smart City Sensing and Communication Sub-Infrastructure

This paper discusses the design and implementation of a smart city sub-infrastructure that provides sensing and communication capabilities in regular and emergency modes. The proposed Smart Box nodes incorporate sensors, a hybrid energy harvester, and a software-defined radio to create a self-powered and self-healing network. The network can aid in disaster management and provide communication services in affected areas.

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/317697320
Smart City Sensing and Communication Sub-Infrastructure
Conference Paper · August 2017
DOI: 10.1109/MWSCAS.2017.8053134
CITATIONS
6
READS
186
6 authors, including:
Some of the authors of this publication are also working on these related projects:
International Network of Disaster StudiesView project
Hadi Habibzadeh
University at Albany, The State University of New York
7 PUBLICATIONS61CITATIONS
SEE PROFILE
Eric Kevin Stern
University at Albany, The State University of New York
44PUBLICATIONS1,473CITATIONS
SEE PROFILE
Tolga Soyata
University at Albany, The State University of New York
79PUBLICATIONS1,810CITATIONS
SEE PROFILE
All content following this page was uploaded by Hadi Habibzadeh on 20 June 2017.
The user has requested enhancement of the downloaded file.

Smart City Sensing and Communication
Sub-Infrastructure
M. Habibzadeh∗
, W. Xiong†, M. Zheleva†, E. K. Stern‡, B. H. Nussbaum‡, T. Soyata∗
∗
Department of Electrical and Computer Engineering, SUNY Albany, Albany, NY 12203
†Department of Computer Science, SUNY Albany, Albany, NY 12203
‡College of Emergency Preparedness, Homeland Security and Cybersecurity, Albany NY 12203
{hhabibzadeh, wxiong, mzheleva, ekstern, bnussbaum, tsoyata}@albany.edu
Abstract—Significant recent research activities and initiatives
by local governments to establish resilient smart city infrastruc-
tures signal that time is right for smart cities in the near future.
For example, sensors deployed within a city could monitor traffic
patterns, perform environmental measurements and determine
optimum traffic routing, when deployed in areas that have a
power infrastructure. In this paper, we conceptualize the deploy-
ment of such nodes, which we term Smart Boxes, in a part of
the city where there is no existing or currently functional energy
and communication infrastructure. We envision our proposed
smart boxes incorporating a multi-source energy harvester (e. g.,
wind/solar). Eliminating the infrastructure requirement allows
our smart box to act as an emergency cell phone network
in any part of the city, thereby forming an emergency sub-
infrastructure. To improve scalability, we use a Software Defined
Radio (SDR) within the box. The contribution of this paper is to
provide an architectural map of the box and a proof-of-concept
experimental demonstration of its LTE network capabilities. Our
experiments show that the box is capable of serving three cellular
users and can be powered from a 50–100 W solar panel and a
50-100 W wind turbine, thereby confirming its feasibility as a
Smart City node.
Index Terms—Smart City, Software Defined Radio, Cellular
Networks, IoT, Autonomous Systems.
I. INTRODUCTION
Recent progress made in conceptualizing resilient smart
city architectures has triggered an avalanche of interest in
municipal authorities and academic disciplines that study both
the technical aspects [1]–[3] as well as the crisis management
and cyber-security aspects of the smart city concept [4]. An ex-
ample is the smart city bill passed by president Obama [5]. En-
visioned developments aim to improve operational efficiency
in a variety of city operations [2], [6] or monitor environmental
conditions [3]. For example, camera-based monitoring of city
traffic can reduce traffic congestion and consequently, air
pollution. In this application, monitoring nodes will be needed
throughout the city that incorporate a camera, multiple air
quality sensors, and an embedded computer [7]. The acquired
data from the sensors and the camera are pre-processed by
the embedded computer and transmitted into the city’s cloud
infrastructure, where an algorithm determines optimum traffic
routing.
The aforementioned applications assume an existing power
infrastructure for the monitoring nodes – an assumption that
often does not hold in the context of disasters and emergen-
cies [8], in which accidents, attacks, or natural hazards may
SMART
BOX
SDR Base
Station
SENSORS
COMPUTER
Energy
Harvester
SMART BOX
SMART
BOX
Wind Solar
WiFi
Direct
Wind Solar
CITY
ACCESS
W S
SMART
BOX
W S
SMART
BOX
WiFi
Direct
Fig. 1. High-level architecture of distributed network of smart boxes. Each
Smart Box is an IoT node and is equipped with a camera and multiple sensors
such as temperature, noise level, NO 2, and O 3. They can harvest energy from
multiple energy sources, such as wind and solar, to operate maintenance free.
They are able to serve multiple cell phone users using a Software Defined
Radio (SDR) and do not require an existing LTE carrier; they can also reach
neighboring boxes via other users’ cell phones.
cause critical power and communication systems to fail [9].
Furthermore, a high availability network connection to the
city is assumed either via wired or wireless networks. Un-
fortunately, such assumptions inherently limit the populations
that can be served by smart infrastructures and often leave
unserved the most vulnerable residents. To bridge this gap,
we conceptualize and prototype a smart city emergency sub-
infrastructure that is self-powered, self-healing, and indepen-
dent of existing infrastructure. At the heart of our design is
our smart city node dubbed Smart Box. These boxes connect
to each other and the Internet to form an independent, yet
globally-connected overlay network for smart and connected
cities. Our Smart Box has two modes of operation – (i) regular
and (ii) emergency. In regular mode, the Smart Box performs
a rich variety of routine monitoring functions such as air
quality control [1], noise pollution detection [3], and traffic
monitoring [2]. In emergency mode, the Smart Box transforms
into a self-configuring and self-healing LTE network that
beyond urban sensing, can provide communication services
in the face of a disaster.

In both operational modes, energy harvesting from multiple
energy sources continues [10], [11], although the operational
priority of the box shifts towards providing a cellular net-
work for the residents affected by the disaster. One of the
most important aspects of the box is easy deployment and
low-maintenance, to enable the city to deploy a significant
amount of them to improve coverage. This poses non-trivial
research/design challenges for the realization of the box.
Our proposed architecture integrates open hardware (such as
SDRs and general purpose PCs) and open-source software to
implement reliable and flexible sensing and communication
infrastructure. Furthermore, we envision the box to incorporate
advanced energy-aware computation algorithms [12] to avoid
energy starvation during emergency scenarios when power is
needed the most.
The contribution of this paper is to prototype and evaluate
Smart Box. In Section II, we provide the technical/architectural
details of our proposed Smart Box. In Section III, we provide
Smart Box operational details that allow these boxes to form
an overlay network sub-infrastructure. We evaluate our design
with a preliminary study using an open source implementation
of the LTE cellular stack called OpenAirInterface and provide
concluding remarks in Section IV.
II. INFRASTRUCTURE DESIGN CONSIDERATIONS
In this section, we provide key Smart Box design consider-
ations and detail the energy harvester. Because the design of
the box and the overlay network – that is composed of many
boxes – are inextricably intertwined, the operational details of
the box in regular and emergency modes are provided within
the context of the network architecture in Section III.
Autonomous Energy Harvesting Capability: Because
the network availability is a crucial operational concern for
the Smart Box, energy harvesting capability is necessary to
ensure operation in locations that lack a power infrastructure.
While solar energy harvesting is a viable alternative to sustain
operations [10], harvesting form multiple energy sources such
as solar and wind provides energy redundancy due to the
complementary nature of these two energy resources, thereby
improving the functional robustness of the box [11].
Cyber Attack Resistant Operation: Undoubtedly, one of
the biggest concerns in the design of the Smart Box is ensuring
its resistance to Cyber attacks. These attacks can be either
system-level or crypto-level; in either case, they attempt to
either disrupt the operation of the box or take over control [13].
Easy Deployment: The density of the deployed boxes in a
localize area has a drastic effect on the quality of the service it
provides in the emergency mode, therefore, these boxes must
be inexpensive and should not require technical expertise to be
deployed. Their solar and wind panels should be fairly easy to
install; ideally this hybrid energy source should be constructed
as single unit with a horizontal wind turbine and a solar panel
on top.
Self-configuration: As the density of the deployed smart
boxes increases, the concern of controlablity arises. We pro-
pose to have a software architecture in each box that incor-
TABLE I
Smart Box FUNCTIONS DEPEND ON ITS OPERATION AND POWER MODES .
Energy/Operation Regular Emergency
Strong Sensing, Backhaul Sensing, Last mile, Backhaul
Normal Sensing, Backhaul Sensing, Last mile
Weak Sensing Sensing or Last mile
porates self-check and self-configuration features. Each box
should be controllable and accessible by city employees and
should be able to accept new firmware revisions through
software uploads. This will enable instant operational improve-
ments of network and sensory capabilities, which are mostly
controlled by software-based components to begin with.
Maintenance-free Operation: While easy deployment re-
duces the initial cost, recurring expenses must also be reduced
to limit long-term costs. Incorporating hardware redundancy
introduces the self-healing feature into the box; continuous op-
eration must not be interrupted by small hardware failures that
are expected in the proposed environments. Functionality of
the box should degrade gradually with each failing component.
III. SMART BOX OVERLAY NETWORK
Our overlay network leverages the Smart Box to create a
blanket coverage for a smart city. The purpose of this network
in regular operation is to provide a sensing and control envi-
ronment. In disasters, the network transitions from a passive
sensing infrastructure to a hybrid sensing and communication
infrastructure that can both aid in determining the severity of a
disaster and in helping the disaster relief efforts by providing
a communication infrastructure in the affected areas. In this
section, we first provide details on our Smart Box design,
components of which are shown in Fig.1. The box features a
set of sensors, a general purpose processor, a hybrid harvester,
a software-defined radio and a suite of open-source software
used for sensor control, data analytics and cross-functionality
switch in the two operation modes. We then detail how our
Smart Box nodes are interconnected in an overlay smart city
network.
A. Smart Box design and implementation
Our Smart Box maintains three key functions: (i) sensing to
collect environmental data, (ii) last-mile communications to
provide user connectivity and (iii) backhaul for interconnect
across Smart Box devices. These functions are prioritized in
importance depending on the operation mode of the node
(regular or emergency) and its energy level (strong, normal
or weak) as detailed in Table I.
In regular mode, the Smart Box performs a variety of
environment monitoring functions. Commercially-available in-
expensive air quality sensors that measure NO 2, CO, CO2,
and O 3 can aid air pollution monitoring. For example, NO2
pollutant is introduced by combustion and can be an indicator
for the degradation in air quality caused by traffic congestion.
Smart Box is designed with heterogeneous sensor capabili-
ties; we envision that deployment of our Smart Box reflects
the sensing and communication diversity of common smart

city applications [7]. Consequently, assuming that different
hardware revisions of the box can incorporate a variety of
evolving sensors, the capability of the software to adapt to
this heterogeneity via the capability of each box to register
itself in the global network is necessary. We propose that
boxes “cache” and know the capabilities of their neighbors
through continuous communication, either with the city or by
one another.
Of particular importance to our system is its ability to
sense failure in existing communication infrastructure and
augment the communication services in that area. To this end,
we harness our previous methodology [14] that monitors the
control channels of commercial networks in order to determine
the degree of congestion at a base station or a backhaul failure.
Additionally, the boxes can detect outages in city lighting
using simple photo-detectors or power-grid outages using Hall
sensors that measure current flow through power lines.
In emergency mode, Smart Box provides combined sensing
and communication functions. Each Smart Box node will fea-
ture functionality to determine the occurrence of a disaster in
its immediate vicinity. Where indications of a disaster scenario
(e.g., power outages, communications network disruptions,
impacts of extreme weather) are detected, the node will notify
the center (denoted “City Access” in Fig. 1) and will transition
to emergency mode. Depending on whether the disaster affects
existing communication infrastructure or not, it will activate
the Smart Box LTE base station capabilities. Activation of the
LTE base station will launch the SDR and will execute the
associated software to spawn the LTE base station.
Our system makes use of an open source implementation
of LTE dubbed OpenAirInterface (OAI) [15]. OAI employs
an SDR, such as the USRP B210, as a radio front-end.
Simultaneously, it runs an eNodeB and an Evolved Packet
Core (EPC) network on a commodity computer in order to
evoke both the base station as well as the LTE core. OAI
implements the 3GPP-defined LTE standards and thus allows
users to access Smart Box for communication services with
their existing unmodified cell phones. A key advantage of
the OAI setup is its modularity. Specifically, an EPC can run
collocated with the base station, virtually creating a mobile
network in a box. This network can either serve its users in
isolation or perform a centralized coordination across multiple
EPCs in order to establish a multi-base-station network. This
enables high resilience of communication services, whereby
in the worst case, the network can provide real-time services
within a cell and delay-tolerant services across cells. Where
cross-cell connectivity is available, the Smart Box network can
provide a wide-area communication service.
B. Energy harvester
For the Smart Box to provide an uninterrupted operation
during unpredictable and harsh environmental conditions, ca-
pability to harvest energy from multiple complimentary energy
sources is necessary. For example, having only solar harvesting
capability will not allow the box to operate at night, unless it
has a large energy buffer. This buffer can be implemented with
rechargeable batteries or supercapacitors [10], both of which
have advantages and disadvantages [16]. While supercapac-
itors have substantially higher life expectancy, rechargeable
batteries embody nearly an order-of-magnitude higher energy
density, thereby allowing the box to be built on a smaller foot-
print. Harvesting from multiple complementary energy sources
provides a steady level of supply, reducing the dependence
on size of the buffer [11]. Therefore, a Smart Box has to
incorporate a multi-source energy harvester, e.g., solar/wind.
C. Overlay Network/Software Architecture
As shown in Fig. 1, the network of Smart Box nodes are
designed to operate individually or as part of an overarch-
ing network. We envision a combination of real-time and
delay-tolerant backhaul to establish the links that intercon-
nect our Smart Box nodes. The real-time interconnect will
be established through long-distance microwave links. The
delay-tolerant backhaul will be leveraged when the real-time
backhaul fails, which can happen if infrastructure is failing
or unable to provision enough energy to sustain its backhaul
links. This delay-tolerant backhaul will be established in a
device-to-device fashion between users phones using Wi-Fi
Direct1. As user move, they will haul information across cells.
The sensor network will connect to a centralized processing
unit controlled by authorized city personnel. This software is
proposed to run in the cloud that is owned – or rented – by
the city. In addition to providing a user interface to the city
personnel, the cloud software also manages the automation
of sensory data gathering by accessing individual Smart Box
data and aggregating it. This data is fed into algorithms that
predict anomalies in traffic conditions, weather conditions, or
environmental conditions. In regular operation, the data can be
used as environmental information, whereas when a disaster is
detected, the surrounding boxes must be alerted to switch their
operation mode to emergency. These algorithms must take into
account the fact that the availability of individual boxes cannot
be guaranteed.
IV. EXPERIMENTAL EVALUATION
In this section we evaluate Smart Box. We are particularly
interested in understanding whether the box can serve multiple
cellular users and whether our hybrid energy harvester can
power the node. To evaluate our setup, we use the aforemen-
tioned OpenAirInterface (OAI) [15]. The eNodeB portion of
OAI runs on a Core i7-6700K based PC with Ubuntu 14.04
(Kernel 3.19) OS and 16 GB of RAM. The EPC portion runs
on a Lenovo Thinkpad X250. We use USRP B210 as a radio
front-end and a Nexus 5 smartphone with Android 5.1.1. as
an LTE client.
We envision an i7-6700K motherboard and a power supply
to be a part of our Smart Box construction, although the
power supply can be replaced with a DC harvester in a field
setting. Therefore, our reported power consumption levels are
conservative. We use a P3 Kill-A-Watt device to measure in-
stantaneous power consumption of the system once per second
1http://www.wi-fi.org/discover-wi-fi/wi-fi-direct

0 2 4 6 8
53.5
54
54.5
55
55.5
Bandwidth (Mbps)
Downlink
Uplink
Fig. 2. Average power consumption vs. bandwidth of a single device that
uploads/downloads data to/from the base station.
and average these samples over two minutes. We evaluate the i)
energy consumption of the entire system without SDR and ii)
the incremental power consumption incurred due to the LTE
functionality including the SDR. We disable all peripherals
such as the discrete graphics card, all unused SATA devices,
Intel Turbo Boost, and the X Server.
Figure 2 shows the processing node power consumption as
the client bandwidth grows from 0 to 8 Mbps with a single cel-
lular user. The idle system power consumption is 47.24 W and
increases by another 6.36–8.26 W, depending on the load. We
observe that uplink is more power-hungry than downlink; as
user demand grows to 8 Mbps, power consumption increases
by 0.9 W for downlink and 1.9 W for uplink.
We further study the effects of increasing number of users
on energy consumption. The current implementation of OAI
supports up to 4 simultaneous users and we conduct our
experiments with a maximum of 3 users. We measure the
power consumption with increasing number of users that
require 4 Mbps each. Figure 3 presents our results. Adding two
more users incurs a 0.82 vs. 2.74 W increase for downlink vs.
uplink power, a pattern that is similar to the previous figure.
V. CONCLUSIONS
In what follows from Fig. 2 and Fig. 3, we conclude that
a Smart Box harvester with a 50–60 W power output can
realize our concept. A proper solar panel and wind turbine
provisioning for this scenario can power this node using a 50–
100 W solar panel and a 50–100 W wind generator [10], both
of which have a manageable physical size for deployment [12].
For future work, we claim that the system power consump-
tion could be drastically reduced by two design changes: i)
the only required DC voltage levels for a PC motherboard
are ± 12 V, ± 5 V, and ± 3 V and can be supplied from a
solar/wind harvester [10], thereby eliminating the 10–20%
inefficiency incurred by the AC-DC conversion of the PC
power supply. ii) in both single- and multi-user experiments,
we observe that the CPU load is only ≈ 20%. Such workloads
can be run power-efficiently on modern CPUs, such as the
14 nm Intel Atom X5-E3940 @9.5 W Thermal Design Point
(TDP), as opposed the i7-6700K @91 W TDP. These improve-
ments will reduce the required solar panel and wind generator
capacities, and consequently their size.
ACKNOWLEDGMENT
This work was supported in part by the National Science
Foundation grant CNS-1239423.
1 2 3
54
55
56
57
58
Number of Devices
Downlink
Uplink
Fig. 3. Average power consumption when multiple nodes, each transferring
data at 4 Mbps, are connected to the eNodeB.
REFERENCES
[1] A. Al-Ali, I. Zualkernan, and F. Aloul, “A mobile gprs-sensors array for
air pollution monitoring,” IEEE Sensors Journal, vol. 10, no. 10, pp.
1666–1671, 2010.
[2] V. Gadepally, A. Krishnamurthy, and U. Ozguner, “A framework for
estimating driver decisions near intersections,” IEEE Transactions on
Intelligent Transportation Systems, vol. 15, no. 2, pp. 637–646, 2014.
[3] N. Maisonneuve, M. Stevens, M. E. Niessen, P. Hanappe, and L. Steels,
“Citizen noise pollution monitoring,” in Proceedings of the 10th Annual
International Conference on Digital Government Research: Social Net-
works: Making Connections between Citizens, Data and Government.
Digital Government Society of North America, 2009, pp. 96–103.
[4] L. K. Comfort, A. Boin, and C. C. Demchak, Designing resilience:
Preparing for extreme events. University of Pittsburgh Pre, 2010.
[5] “$80 million in new federal investment and a doubling of
participating communities in the white house smart cities
initiative,” https://www.whitehouse.gov/the-press-office/2016/09/26/
fact-sheet-announcing-over-80-million-new-federal-investment-and.
[6] J. Jin, J. Gubbi, S. Marusic, and M. Palaniswami, “An information
framework for creating a smart city through internet of things,” IEEE
Internet of Things Journal, vol. 1, no. 2, pp. 112–121, 2014.
[7] H. Habibzadeh, Z. Qin, T. Soyata, and B. Kantarci, “Large Scale Dis-
tributed Dedicated- and Non-Dedicated Smart City Sensing Systems,”
IEEE Sensors Journal, 2017, accepted for publication.
[8] C. F. Parker, E. K. Stern, E. Paglia, and C. Brown, “Preventable catastro-
phe? the hurricane katrina disaster revisited,” Journal of Contingencies
and Crisis Management, vol. 17, no. 4, pp. 206–220, 2009.
[9] E. Stern and L. Svedin, Auckland unplugged: Coping with critical
infrastructure failure. Lexington Books, 2003.
[10] M. Hassanalieragh, T. Soyata, A. Nadeau, and G. Sharma, “UR-
SolarCap: An Open Source Intelligent Auto-Wakeup Solar Energy
Harvesting System for Supercapacitor Based Energy Buffering,” IEEE
Access, vol. 4, pp. 542–557, Mar 2016.
[11] M. Habibzadeh, M. Hassanalieragh, A. Ishikawa, T. Soyata, and
G. Sharma, “Hybrid Solar-Wind Energy Harvesting for Embedded
Applications: Supercapacitor-based System Architectures and Design
Tradeoffs,” IEEE Circuits and Systems Magazine, 2018, accepted for
publication.
[12] M. Zhu, M. Hassanalieragh, A. Fahad, Z. Chen, T. Soyata, and K. Shen,
“Supercapacitor Energy Buffering for Self-Sustainable, Continuous
Sensing Systems,” University of Rochester, Department of Computer
Science, Tech. Rep. TR–995, Mar 2016.
[13] O. Kocabas, T. Soyata, and M. K. Aktas, “Emerging Security Mecha-
nisms for Medical Cyber Physical Systems,” IEEE/ACM Transactions
on Computational Biology and Bioinformatics (TCBB), vol. 13, no. 3,
pp. 401–416, Jun 2016.
[14] P. Schmitt, D. Iland, M. Zheleva, and E. Belding, “Hybridcell: Cellular
connectivity on the fringes with demand-driven local cells,” ser. INFO-
COM16, San Francisco, CA, 2016.
[15] N. Nikaein, M. K. Marina, S. Manickam, A. Dawson, R. Knopp, and
C. Bonnet, “OpenAirInterface: A Flexible Platform for 5G Research,”
SIGCOMM Comput. Commun. Rev., vol. 44, no. 5, pp. 33–38, Oct.
2014. [Online]. Available: http://doi.acm.org/10.1145/2677046.2677053
[16] G. Honan, N. Gekakis, M. Hassanalieragh, A. Nadeau, G. Sharma,
and T. Soyata, “Energy Harvesting and Buffering for Cyber Physical
Systems: A Review,” in Cyber Physical Systems - A Computational
Perspective. CRC, Dec 2015, ch. 7, pp. 191–218.
View publication statsView publication stats