Professional Documents
Culture Documents
Smart Street Lamp System Using Lorawan and Artificial Intelligence Part I
Smart Street Lamp System Using Lorawan and Artificial Intelligence Part I
Abstract— This paper presents the design of a smart several kilo bytes per second) and consume low
street lamp system around a football field at King power [2]. LoRaWAN is therefore often referred to
Mongkut's University of Technology Thonburi as the best technology in the Internet of Things (IoTs)
(KMUTT). The system utilized the LoRaWAN wireless applications where low speed remote communication,
communication protocol transmit at a distance over 500
low energy, high link budget, and superior resistance
meters between multiple LoRa nodes (Heltec Cubecell
HTCC-AB01 boards) and a LoRaWAN gateway. By to interference is required [3].
combining LoRaWAN, cloud computing server and According to the World Energy Outlook
Artificial Intelligence technologies. The smart street organization, overall energy demand is set to drop by
light system is able to detect 4 object classes (pedestrian, 5% and as much as 10% in some regions globally in
bicycle, motorbike, and vehicle) and control street the year 2020 due to Covid-19. However, there are
lamps around the KMUTT football field at night. some exception to this forecast including Thailand
Multiple IP cameras were place around the football where the country’s energy consumption is rising by
fields and a Nvidia Jetson nano AI board process 1.8% according to the Energy Policy and Planning
images information. When AI is detected at least one
Office (EPPO). With increase energy demands, the
object class, the smart street lamp will increase the
lamp’s brightness to 100%. If no object is detected for a need to save energy is more important than ever. By
period of time, the lamp’s brightness decreases 50% to deploying LoRaWAN wireless technology in
save the energy unless more object is detected. conjunction with Artificial Intelligence (AI) image
processing algorithm, smart street lamps, an energy
saving system is presented in this paper. The system
Keywords— LoRaWAN, AI, Cloud Computing, Smart consisted of multiple LoRaWAN nodes, a LoRaWAN
Street Lamp, IP Camera, Nvidia Jetson Nano, KMUTT gateway, an AI processing unit (NVidia Jetson
Nano), IP cameras, a Node-red software and a Cloud
I. INTRODUCTION computing server.
WiFi and Bluetooth wireless communication It is noted that this article will be focusing on the
technologies are very popular since they are LoRaWAN communication datalink and the street
ubiquitous and easy to use. One drawback for these lamps brightness control. The next paper (part II) will
communication technologies however is the focused on the intelligence (AI) – image processing,
communication range [1]. LoRaWAN or Long- the underline decision making algorithm and the
Range Wide Area Network is a wireless cloud computing server software process.
communication technology that consume low energy
while offering extremely long range when compare to II. System Description
many other wireless technologies such as WiFi, A. System Overview
Bluetooth, Zigbee, Z-Wave, LTE-M and NB-IoT.
The system overview is summarized by Fig. 1
With these advantages, LoRaWAN is perfectly
below and the LoRaWAN node deployed in this
suitable for long distance remote wireless
project is the Heltec Cubecell HTCC-AB01 board (as
communications (many kilometers) that do not
shown in Fig. 2) attached to the street lamp driver
require high speed communication (data rate up to
circuit to control the lamp’s brightness. These street
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on July 01,2021 at 04:05:29 UTC from IEEE Xplore. Restrictions apply.
2021 International Electrical Engineering Congress (iEECON2021)
March 10-12, 2021, Pattaya, THAILAND
lamps were installed around the university’s football It can be seen that various information as show
field. Each LoRaWAN node continuously listening in Fig. 3 must be included in the arduino code before
for commands from the LoRaWAN gateway located it can be programmed into the LoRaWAN nodes.
on top of the class room building 4 (CB4) at a Unique parameters such as deviceEUI, AppEUI and
distance of approximately 500 meters away. AppKEY are require for the node to communicate
with the gateway [4]. The same deviceEUI, AppEUI
and AppKEY must be preprogrammed into the
gateway and the nodes. Any mismatch information
will result in both devices will not be able to
communicate to each other. Since the node units are
powered by an AC power supply, the LoRaWAN
class can be set either class A or C. LoRaWAN class
A will consume more energy since both devices
(gateway and nodes) are on all the time. LoRaWAN
class C is essential for a battery operation. Since
LoRaWAN class C require the least power when
compare to the class A and B due to both gateway
and nodes only transmit and receive at a pre-
Fig. 1. System Overview determine period [5].
The system helps save electricity energy by C. Smart Street Light On/Off and Brightness Control
setting the street lamp brightness at 50% when the AI The smart street lamp brightness control is based
image processing detecting no pedestrian, bicycle, on the lamp’s driver, a Bipolar Junction Transistor
motorcycle and vehicle around IP camera looked out (BJT) circuit, and the Pulse Width Modulation
areas. As the camera detected the objects (people, car, (PWM) in the LoRaWAN node. To turn on or off the
bicycle and motorcycle), the LoRaWAN gateway lamp completely, a small 5A relay is used.
transmits a command set to the LoRaWAN node to The smart street lamp’s driver input voltage
brighten the street lamps to 100% for a period of time range from 100-277 V, current rating of 0.33 A, rated
before the lamp’s brightness return to 50% to save power 70 W, and 0.95 power factor. The lamp’s
energy. This cycle continues all night to ensure the controller has 5 connections: L (line), N (Neutron),
pedestrians’ safety and security while saving energy. ground, Dim+ and Dim-. To control the lamp’s
brightness, a BJT driver circuit is needed to connect
B. Setting up LoRa nodes and a Gateway to both Dim+ cable and Dim- cable.
In order for LoRa nodes (Heltec Cubecell HTCC- The Fig. 4 shows an arduino IDE PWM control
AB01 board) and the gateway to communicate to code to be programmed into the LoRa nodes. Using a
each other, an arduino code is needed to be BJT circuit in Fig. 5 to drive the PWM signal on
programmed to the LoRa nodes. Dim+ and Dim- pins.
Fig. 2. LoRa node - Heltec Cubecell HTCC-AB01 Fig. 4. BJT circuits for simulation
To perform calculations for the above BJT circuit
design, based on the calculation equation (1)
VB = VIN – VCE (SAT) (1)
The value VB = 2.6V, and by selecting RB =
10kΩ, IB = 2.6x10-4 A, according to the BJT
datasheet, the bata value β = 250 for the bc337
transistor. Because IC = βIB, the calculated IB = 0.065
A and the value of a resistor at the transistor’s
collector (Rc), Rc = 143.07Ω. In practice, one will
choose the closest resistance available in the market
which is 180Ω.
190
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on July 01,2021 at 04:05:29 UTC from IEEE Xplore. Restrictions apply.
2021 International Electrical Engineering Congress (iEECON2021)
March 10-12, 2021, Pattaya, THAILAND
III. EXPERIMENT
There were five different tests performed before B. LoRaWAN data transmission test
the final system was to be deployed in the field to This test determined how much LoRaWAN data
ensure the system robustness and reliable operations. rate effected by many factors including distance
These testes would include LoRaWAN distance vs between transmitter and receiver, Spreading Factor
Received Signal Strength Indicator (RSSI), data (SF) value, and each node’s Signal to Noise Ratio
transmission, lamp’s on/off control, lamp’s brightness (SNR). Fig. 7. and Fig. 8. revealed the RSSI, SNR and
control, and the Node-Red software control system data rate at the gateway.
integration tests.
A. LoRaWAN Distance vs Received Signal Strength
Indicator (RSSI)
It could be observed that radio signal strength
depends on distance between the transmitter and the
receiver. The signal strength of the receive decreases
as the distance from the transmitter increases. This
phenomenon obeys the “inverse square law” of radio
propagation. In this experiment, a transmitter
(gateway) was kept stationary and receivers (nodes) Fig. 7. LoRa Server webpage
moved to different locations in the KMUTT campus.
The results showed that the RSSI decreased as the
distance increased as shown in the Fig. 5.
191
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on July 01,2021 at 04:05:29 UTC from IEEE Xplore. Restrictions apply.
2021 International Electrical Engineering Congress (iEECON2021)
March 10-12, 2021, Pattaya, THAILAND
192
Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on July 01,2021 at 04:05:29 UTC from IEEE Xplore. Restrictions apply.