LoRa
Technology
3.1 Introduction
The LoRa network relies on two key components, LoRa and LoRaWAN, where LoRa presents
the physical layer modulation whereas the LoRaWAN represents the MAC layer protocol.
Combining these two components results in a low power and cost-effective wide-area network.
We discuss the PHY and MAC layer issues separately.
3.2 Physical layer
3.2.1 Propagation models
Accurate prediction of LoRa signal behavior in the intended deployment environment is
essential for effective network planning. This enables network designers to estimate the coverage
of end devices (EDs) and determine the appropriate number of gateways (GWs) required. To
achieve this, propagation models are widely utilized, as they offer a reliable approach for
estimating received signal strength by calculating path loss. These models can generally be
classified into three main types: (i) empirical, (ii) deterministic, and (iii) stochastic.
Over the years, numerous wireless propagation models have been developed, primarily
originating from three main domains: standardized bodies, industry providers (vendors or
operators), and academic research. Among these, the Okumura-Hata model is frequently applied
within LoRaWAN contexts. The Hata model, in particular, provides a set of equations derived
from the original Okumura framework, adjusting various parameters to suit different scenarios.
These models are instrumental in conducting simulation-based analyses of signal propagation,
considering factors such as frequency range (150–1500 MHz), communication distance (1–20
km), and antenna elevation.
An extension of the Hata model, known as the COST 231-Hata model, broadens the applicable
frequency band up to 2000 MHz. This variant supports predictions for systems operating
between 1500 and 2000 MHz, with communication distances spanning from 1 to 20 km. It also
accommodates a range of antenna heights—1 to 10 meters for EDs and 30 to 200 meters for
GWs. Additionally, the COST-231 Walfish-Ikegami model (often abbreviated as COST-WI)
integrates features from both the Walfish and Ikegami models. It enhances path loss estimation
by incorporating environmental factors such as street width, building height, and spacing,
making it well-suited for modeling urban areas of varying density.
3.2.2 Signal-to-Noise Ratio (SNR) and Receiver Sensitivity
To successfully decode a signal, a LoRa receiver must achieve a minimum signal-to-noise ratio
(SNR). This SNR threshold varies with the spreading factor (SF). Receiver sensitivity S is
determined by:
S = -174 + 10 log10 (BW) + NF + SNRth
• BW: Bandwidth (typically 125, 250, or 500 kHz)
• NF: Noise figure (usually 6 dB)
• SNRth: Threshold SNR for reliable demodulation, varies per SF
Spreading Chips /Symbol SNR limit Data Rate
factor
7 128 -7.5 db 5469 bps
8 256 -10 3125
9 512 -12.5 1758
0 1024 -15 977
11 2048 -17.5 537
12 4096 -20 293
Table3.1 CORRESPONDING SPREADING FACTOR FOR EACH CHIP/SYMBOL, SNR LIMIT, AND DATA
RATE FOR LORA COMMUNICATION OVER A CHANNEL OF 125 KHZ BW
3.2.3 Data Rate and Modulation
LoRa uses Chirp Spread Spectrum (CSS) modulation, which provides robustness against
interference. The data rate depends on spreading factor (SF), bandwidth (BW), and coding rate
(CR):
DR = {SF*BW} \ 2sf * CR
• Higher SF increases transmission range but lowers data rate.
• Higher BW increases data rate but reduces range due to lower sensitivity.
fig 3.1 Data rate versus spreading factor for CR = 4/5.
3.2.4 LoRa Frame Structure
Each LoRa packet consists of:
• Preamble: Synchronizes receiver and transmitter
• Optional Header: Contains payload length, coding rate, etc.
• Payload: Carries application or MAC layer data
• Optional CRC: For error detection.
Fig 3.2 LoRa frame structure.
3.2.5 Chirp Signal Modulation
LoRa encodes data using chirps—frequency-swept signals. There are:
• Up-chirps: Frequency increases over time
• Down-chirps: Frequency decreases over time
• SF: Determines how many bits per symbol (e.g., SF=7 encodes 7 bits/symbol)
Fig 3.3 LoRa chirp signal
3.2.6 Coding and Interleaving
LoRa uses Hamming codes for forward error correction (FEC), followed by:
• Whitening: Randomizes bit patterns to avoid burst errors
• Interleaving: Spreads symbols across time
• Gray coding: Reduces bit error rate
Fig 3.4 LoRa channel coding diagram.
3.2.7 Time on Air and Spreading Factor
In LoRa communication, the Time on Air (ToA) of a packet is a critical parameter influenced
primarily by the Spreading Factor (SF). Time on Air denotes the total duration a packet occupies
the channel and is determined by SF, bandwidth (BW), coding rate (CR), and payload size.
Mathematically, higher SF values increase the number of chirps required per symbol, thereby
extending the symbol duration and overall transmission time. This makes SF a central element in
balancing data rate, transmission range, and energy consumption.
While higher SF values (e.g., SF12) offer greater sensitivity and thus longer communication
ranges, they significantly increase the ToA. This leads to increased channel occupancy and
higher susceptibility to collisions in dense deployments. Conversely, lower SFs result in shorter
ToA but at the cost of reduced range and robustness. The formula for ToA considers both
preamble and payload duration, making it a useful metric in estimating battery life and ensuring
compliance with regional duty cycle regulations.
3.2.8 Frequency Bands and Regional Considerations
LoRa operates in various sub-GHz unlicensed ISM bands, which differ based on geographic
regions. Common frequency bands include:
• 868 MHz (Europe)
• 915 MHz (North America)
• 433 MHz (Asia and some European countries)
These bands are subject to different regulatory policies, affecting transmission power limits, duty
cycles, and channel access. For instance, the ETSI EN300.220 regulation in Europe enforces a
1% duty cycle on devices operating in the 868 MHz band, whereas FCC Part 15 in the United
States allows more leniency in the 915 MHz band by permitting techniques such as frequency
hopping.
3.2.9 Transmit Power and Duty Cycle Limitations
Transmit power is another key parameter in the physical layer, influencing both the link budget
and the range of communication. LoRa devices generally transmit at power levels ranging from
+14 dBm to +20 dBm, depending on the region and hardware capabilities.
However, regulatory limits imposed by authorities such as ETSI and FCC restrict the maximum
power and transmission time. In Europe, for example, LoRa devices operating in the 868 MHz
band are subject to a 1% duty cycle, meaning a device can only transmit for 36 seconds per hour
per channel. This limitation is intended to ensure fair spectrum sharing among multiple users and
technologies within the ISM band.
To adhere to these constraints, network planning must account for duty cycle restrictions,
especially in high-traffic or latency-sensitive applications. Techniques such as channel hopping,
payload optimization, and ADR (Adaptive Data Rate) can help mitigate the impact of these
restrictions.
3.2.10 Link Budget
The LoRa technology has a link budget greater than any other existing technologies. Link budget
accounts for its long range to a great extent.
Link budget is an accounting of all the gains and losses in a transmission system. The link budget
of a network can be expressed as
PRX(dBm) = PTX(dBm) + GSYSTEM(dB) - LSYSTEM(dB) - LCHANNEL(dB) - M(dB)
where,
PRX(dBm): received power
PTX(dBm): transmitted power
GSYSTEM(dB): system gains such as those associated with directional antennas, etc.
LSYSTEM: losses associated with the system such as feed-lines, antennas etc.
LCHANNEL: losses due to the propagation channel
M: fading margin
The link budget is the power that is received at the receiver side. The link budget of LoRa is high
that in turn accounts for its high sensitivity. Most of the technologies used for connectivity in IoT
uses frequency shift key (FSK) modulation. When the data rate of LoRa signal is equivalent to
four times the data rate of the frequency shift key (FSK) signal the LoRa signal offers the similar
or equal sensitivity. Thus, LoRa can cover more distance than any other technique.
3.3 NETWORK ARCHITECTURE
LoRa is a physical layer that provides a long-range communication link. This has been
standardized and extended by adding a MAC layer LoRaWAN (Long Range Wide Area
Network) to it. This defines the network architecture and communication protocol. LoRaWAN
specification is standardized and open sourced by the LoRa alliance.
Most of the existing technologies are based on mesh network. In the mesh network the
infrastructure node is connected to as many nodes as possible and cooperates with one another to
route the data. In the mesh network each node receives and forward data from other node that
might be irrelevant for it. This increases the range to a great deal but also adds complexity and
decreases the battery life. On the other hand, in the star topology the bridge or switches are
directly connected to a small subset of bridge or switches deceasing complexity of the network.
These provide a hierarchical infrastructure. LoRaWAN is based on star topology. They decrease
the power consumption and battery life to a great extend in comparison with the conventional
mesh network.
The LoRa network consist of four basic elements
• The LoRa node or End Points
• The Gateway
• The Network Server
• Application Server
Fig3.5 LoRaWAN network architecture
3.3.1 The LoRa Node or End Points
The end points or LoRa nodes comprises sensors or application where sensing and control takes
place such as sensors, tracking devices, etc., [1] [4]. This is the heart of an embedded system.
These include water meter, smoke alarm, gas monitoring and such applications.
3.3.2 The Gateway
The gateway comprises the net element of the LoRa network. There are a number of gateways .
Each gateway is connected to every end node. The data transmitted by the node is sent to all the
gateways and each gateway which receives a signal transmits it to a cloud based network server
via cellular, Ethernet, satellite, or Wi-Fi. The gateway can be micro gateway or pico gateway.
Micro gateways are used in public network to give city or nationwide coverage while the Pico
gateway is used in hard to reach dense areas to improve the quality of service and network
capacity. Micro gateway gives high coverage. Both Omni-directional and multi-sector antennas
are used in LoRa base station.
3.3.3 The Network Server
The network server has all the intelligence. The data received from different gateways is filtered,
security checks, adaptive data rates etc., are performed and acknowledgement is sent to the
gateways . The network server is the one who identifies whether the data received is intended
for any application server and is thus sent to the intended application server via some backhaul.
3.3.4 The Application Server
They receive the intended data from the network server.
3.4 LoRaWan Features
3.4.1 Battery lifetime
The most important criteria of an embedded device is its battery lifetime. Most of the embedded
devices need to communicate with other devices near or far. This consumes high power. The
embedded devices are all mostly battery operated. Thus, the essential requirement of these
embedded devices is its battery lifetime. Most of the protocols or techniques used to create IoT
embedded device now-a-days consumes very high power thus reducing the battery lifetime.
LoRa optimizes the battery consumption in a device and is most suited for battery operated
embedded device. LoRa consumes the least power when compared to all the existing
technologies.
The low battery consumption in a LoRa networks accounts to the asynchronous communication
of the nodes in the network. In a LoRa network the nodes communicates only when they have
any data to send whether energy driven or scheduled. They adopt the ALOHA method. In
ALOHA method frame is send only when there is any data to send otherwise no transmission
takes place. If the frame is received successfully another frame is sent or the same frame is
retransmitted. The ALOHA method saves battery lifetime because the network is doing
transmission when there is any data. It does not do any transmission otherwise. Also, most of the
other technologies are of mesh network or adopt synchronous communication in which the nodes
have to wake up and synchronize time to time. This consumes more energy. In a recent study it
was shown that the LoRa technology is 4 to 5 times better than any other existing technology.
3.4.2 Network Capacity
Star network is used in Lora network. The LoRa network gateway receives data from large
number of nodes. For this the gateway must have high capacity. This is achieved by adaptive
data rate and multichannel multi-modem transceiver at the gateway.
Adaptive Data Rate (ADR) is a mechanism for
optimizing data rates, airtime and energy consumption in the network. Static nodes use ADR
(Adaptive Data Rate)[6]. Whether to use ADR or not is decided by the node itself. In ADR the
data rates of the node are managed by the network. When the node decides that it wants to use
ADR it sets the ADR bit to 1 in the uplink transmission. When the network gets the signal that
the node wants to use ADR it collects the signal to noise ratio, data rate, number of gateways that
receive the data and signal strength of the 20-recent transmission from the specific node. Based
on this data the network decides how much it can increase the data rate of decrease the
transmission power. This reduces the air time and optimizes the transmission power. We can also
say that it helps in reduction of battery consumption. This whole thing is done using the ADR
algorithm. A node that is static at one time and mobile at another. So, the ADR algorithm can
also be used by a mobile node that is parked at a fixed spot at the
given time [13].
The multichannel multi-modem transceiver enables messages to be simultaneously received from
multiple gateways.
3.4.3 Security
AES encryption and IEEE 802.15.4/2006 Annex B is used in this technology for security and
authentication. While most technologies incorporate single layer security, LoRa network
incorporate two layers of security: network security and application security. The network
security is used for authenticating the node in the network while the application security protects
the end user application data from the network operator. The LoRa technology uses two keys for
the security and authenticity: NwkSKey (Network Session Key) and AppSKey (Application
Session Key).
For the end device to participate in a network it must be activated and authenticated. The
technology has two methods of authentication and activation
1. Over the air activation (OAA)
2. Activation by personalisation (ABP)
1) Over the air activation (OAA):In this type of end device activation, the device is not
personalized with any information. A join procedure is done for the end device to join any
network. Before joining the network, the end device is loaded with the information. This then has
to be repeated for every transmission over the network when the session context information is
lost. This method ensures that the end devices is not confined to any particular service provider
and can join any network service provider while in roaming.
2) Activation by personalization (ABP):In this type of activation, the end device will be already
stored with the information needed for the activation. The device directly joins the specific
network defined in the information when the device is started.
This type of personalization is not used frequently and is used only under some specific cases.
The commonly used personalization method is the Over the Air Activation method. In ABP
method simply a join request and join accept message is sent between the end device and the
network for the end device to activate. NetSKey and AppSKey must be specific to every end
device. Apart from being an activation method both the methods also provide authenticity and
security over the network.
3.4.4 Quality of Service ( QoS )
Quality of service is the account of overall performance of a network. It is based on various
parameters such as data rate, immunity, throughput, packet loss, etc. The LoRa technology which
is based on chirp spread spectrum technology offers fairly good quality of service. It is immune
to interference, multipath and fading.
In a wireless network, as the distance between the devices increases the signal strength
decreases. These are usually avoided by installing repeaters or by having additional nodes as in
mesh topology. But the cost of having more repeaters or nodes in between is very high. But for
LoRa, those signal with different sequence will be treated as noise at the network coordinator.
The nodes near the coordinator can transmit at a higher data rate and ones far away from the
coordinator can reduce the bandwidth.
3.5 MAC PROTOCOL
LoRa defines the physical link layer while LoRaWAN defines the communication protocol and
the network architecture. The end nodes in a LoRa network are divided into three basic classes
according to their battery lifetime and the downlink communication latency [5]. The end device
in a LoRa network has different requirements and serves different applications. According to
these the end devices are put in any of these classes.
The three-basic class of the end devices in LoRa network is
Class A
Class B
Class C
A. Bi-Directional End Devices (Class A)
These are the lowest powered end device system. The devises under this class follows aloha
method of communication. These devices have one uplink transmission slot and two downlink
receive slots.
Fig 3.6 End device receive slot timings
Here first the end device transmits an uplink message and opens up two receive slots. The first
receive slot is opened after a delay of +/- 20 microsecond. The downlink data rate and the
downlink frequency are a function of uplink data rate and uplink frequency. Again, after a delay
of +/- 20 microseconds the second downlink receives window is opened. The data rate and the
frequency in this slot is configurable. They can be configured using the MAC command. Only at
these two receive slots the server can sent data. If the server wants to send more data, it has to
wait for the next uplink transmission from the end device this class can only be used by those
applications that need to send small data and that has to send data only after the end device sent
an uplink.
The duration of the receive window must be long enough for the end device to at least detect the
preamble of the downlink message. Once the end device detects the preamble it stays active until
it receives and demodulates the frame. Data can be received by the end device in any one of the
two receive slots. If the data is received at first receive slot, the end device process the data,
demodulates it and if the data is intended for that end device the end device does not open the
second receive slot. If the network has to do downlink transmission it starts the transmission at
the starting of any of the receive slot. The end device is not ready to receive next downlink
message until and unless it has received a message in any of the two receive slot for the previous
uplink transmission or till the receive slots expires.
B. Bi-directional end-devices with scheduled receive slots (Class B)
In class B addition to the random receive slots of class A there is additional scheduled receive
slots. Therefore, there are more receive slots than class A which opens at scheduled times. The
gateway sent scheduled beacon to the end device for the end device to open their additional
receive slot at periodic times. The server then knows that the end device is listening. The
additional times slots are called ping slots. This class is used when additional receive slots are
needed at predictable time in addition to the slots of class A that is available after an uplink
transmission from the end device to the server.
In this class all the gateways synchronously send the beacon to the end device. The end device
opens up the receive slot called ping slot at the specified time. The network then sends the data to
the end device at the specified slot. The downlink communication is called ping. The network
selects the gateway for the downlink communication depending on the signal strength of the last
update of uplink from the end device. If the end device moves and finds any change in the signal
strength it should notify the network in the uplink transmission. The network then updates this in
its database. While joining every device joins as class A and then shifts to class B.
C. Bi-directional end-devices with maximal receive slots (Class C)
Unlike class A and class B, the class C devices open their receive slots all the time. The receive
slot is closed only when the end device is transmitting. Since the class C end devices have their
receive slot open all the time they consume more power than the other two classes. But these
classes provide lowest latency for communication. These can be used for only those devices that
do not have any power constrains.
Fig 3.7 Class C end device receive slot timings
The class C end devices open a receive slot 2 in between the end device class A transmit and
receive slot 1. When the receive slot 1 is closed the receive slot 2 is again opened until the next
transmit or receive slot. Whenever there is no transmission or reception in the usual class A slots
the receive slot 2 opens up indefinitely until next transmit or receive. The class C end device
always have the receive slot 2 open and thus the end device can receive data from the server.
These Class end devices have the most power requirement than class A or class B end devices.
