Lora-Based Mesh Network For Iot Applications: Heon Huh Jeong Yeol Kim
Lora-Based Mesh Network For Iot Applications: Heon Huh Jeong Yeol Kim
978-1-5386-4980-0/19/$31.00
Authorized ©2019Metropolitano
licensed use limited to: Instituto Tecnologico IEEE 524 on September 22,2021 at 21:40:35 UTC from IEEE Xplore. Restrictions apply.
ITM. Downloaded
2019 IEEE 5th World Forum on Internet of Things (WF-IoT)
server. Unlike a conventional cellular system, there is no initial developed a modified LoRa protocol stack that supports mesh
gateway search procedure at the end-nodes, and each node is networking and time-division multiple-access. Because con-
not associated with a specific gateway [9]. ventional network topologies, such as star-and-tree topologies,
A packet from a particular node is delivered to the network are well established, highly standardized and vendor-neutral,
server through multiple gateways. The network server decodes they are often used for public networks. Common standard and
duplicate packets relayed from the multiple gateways and interoperability are not well-defined for mesh networks, which
chooses the best reliable packet. If necessary, the network are often used for private networks with specific application
server sends back an acknowledgment to the end-node through scenarios [13].
the corresponding gateway. LoRaWAN uplink achieves diver-
A. LoRa Mesh Network in ISM Band
sity gain and hand over for mobile end-nodes implicitly [11].
The working scenario of LoRaWAN assumes the prelimi- Fig. 2 shows the LoRa mesh network architecture. As there
nary condition that there are enough well-planned gateways is no hierarchy in a mesh network, every node can relay a
such that each node can reach one or multiple gateways in packet and cooperate with other nodes to efficiently route a
a single hop of LoRa communication. Optimal cell-planning packet to the gateways. Mesh networks dynamically connect
may not be an issue for public networks run by network end-nodes together and self-configure the routing paths [14].
operators. However, for a temporarily run private ad-hoc In LPWAN-based IoT systems, the end-nodes are low-
network (e.g., in construction fields), the network deployment powered devices and need to operate for many years on a
should be simple, reliable, and scalable. In fact, there are many coin-cell battery. Therefore, the transmission power of the end-
use cases where private networks are preferred over public nodes is usually maintained lower than that of the gateways.
networks that require subscription and monthly fee for every In some countries, different maximum transmission powers
node. are designated for uplink and downlink by regulation. This
In the LoRaWAN topology, direct communication is not imbalance causes a problem in a real mesh network.
supported between the LoRa nodes. Any node-to-node com- Generally, a mesh network can be self-configured using
munication must be through the network server via two- flooding algorithms. Flood-based networking can be summa-
gateway transmission. In LoRaWAN, there is no multiple rized as follows [14]. First, a gateway broadcasts a beacon
access scheme to prevent the collision between the packets message. The end-nodes receive the message and relay it with
coming from the end-nodes, whereas carrier-sense multiple a reception history. When the flooding process ends, most of
access (CSMA) is employed in conventional wireless net- the end-nodes would have received the beacon message several
works, such as 802.11, to mitigate the collision rate. The link times either directly from the gateway or through other end-
capacity of LoRaWAN decreases to that of pure Aloha when a nodes. Thereafter, a mesh table is prepared for each end-node
significant number of node transmissions occur simultaneously based on the reception history and received signal strength
[12]. indication (RSSI) of the beacon message.
In summary, there are two problems when LoRaWAN is Fig. 3 shows the LoRa network with power imbalance [15].
used to serve applications in a poorly cell-planned private The mesh network assumes a maximum of three hops, and
network: 1) Presence of end-nodes that are disconnected the one-hop coverage of the gateway is up to the imbalance
from the network. 2) A high collision rate because of the region denoted by the gray area. To initiate mesh networking,
interference among the end-nodes. a gateway broadcasts a beacon. If a node is located outside the
gateway coverage, it tries to connect with a neighboring node
III. M ODIFIED L O R AWAN for uplink transmission. The end-node, indicated using a red
To address the problems associated with standard Lo- circle, is within the coverage of the gateway and can receive
RaWAN when employed in ad-hoc private networks, we the downlink beacon directly from the gateway. Accordingly,
525 on September 22,2021 at 21:40:35 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: Instituto Tecnologico Metropolitano ITM. Downloaded
2019 IEEE 5th World Forum on Internet of Things (WF-IoT)
the gateway and neighboring end-nodes indicated using the C. Gateway development
other red circles are on top of its mesh table. However, an We developed a gateway for the proposed LoRa protocol.
uplink packet transmitted by the end-node in the imbalance Fig. 4 shows the system architecture of the gateway. Following
region can not be delivered to the gateway neither directly are the advantages of the gateway:
nor through the neighboring nodes indicated using the red • Applicable to most IoT operation scenarios
circles. This imbalance may lead to a time-out or looping in • Supports a maximum of four LoRa channels (end-node
the routing procedure when using the conventional flooding mesh network)
mesh algorithm. • Additional wireless backhaul with 38.4kbps frequency-
To overcome the aforementioned issues, we remove the shift keying (FSK)
gateway and the neighboring end-nodes from the mesh table • Ethernet (WAN connection)
and place the end-node, indicated using a blue circle, on top • Indoor and outdoor use (waterproof / dust proof design)
of the mesh table in our modified LoRa protocol if the current
end-node is in the imbalance region, as shown in Fig. 3. IV. U SE C ASES OF T HE M ODIFIED L O R AWAN
Following are the characteristics of the proposed mesh In this section, we present three use cases of the proposed
network: LoRa protocol, which can be implemented in a private net-
work.
• There is no need for cell-planning with the gateways, and
the end-nodes can be deployed in a relatively simple and A. Fire Pipe Freeze Monitoring System
flexible manner. Fig. 5 shows the deployment site. There are more than 600
• The routing paths are established automatically among fire-pipe temperature sensors in an area covering 2.3km ×
the end-nodes. If there is a failure in the existing routing 4.8km. In an environment such as this, it was difficult to find
path, re-routing is done in the network, and new paths sufficient sites for gateway installation. Because of the severe
are reconfigured. radio interference and channel variation, we reconfigured the
• The network is optimized for low-rate communication at routing path periodically. The proposed LoRa mesh network
sub-GHz ISM bands. worked successfully.
• N:1 and 1:N bidirectional communications are possible.
• The network is suitable for smart metering, building en-
ergy management systems (EMSs), and safety monitoring
in construction fields.
526 on September 22,2021 at 21:40:35 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: Instituto Tecnologico Metropolitano ITM. Downloaded
2019 IEEE 5th World Forum on Internet of Things (WF-IoT)
527 on September 22,2021 at 21:40:35 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: Instituto Tecnologico Metropolitano ITM. Downloaded