1. Introduction
In recent years, the application of the Internet of Things (IoT) in the civilian sector has witnessed rapid growth, encompassing various domains such as agriculture, environmental monitoring, and smart cities. These application areas urgently need wireless technologies that can provide low-cost, scalable, and robust connectivity on a large scale [
1,
2,
3]. The successful deployment of IoT applications depends on seamless connectivity and ensuring the security and integrity of transmitted data in diverse IoT environments [
4,
5]. In this regard, the emergence of Low-Power Wide-Area Networks (LPWAN) precisely caters to the specific requirements of these applications. LoRa, derived from the chirp spread spectrum (CSS) modulation technology developed by Semtech, represents a long-distance communication technology. As one among several LPWAN technologies [
6], LoRa has undergone rapid development and possesses relatively mature technology. Upon its introduction, LoRa garnered significant attention due to its remarkable sensitivity, robust interference resistance, and outstanding network capacity [
7]. Addressing the unique challenges of IoT environments, including the need for secure and reliable data transmission, is paramount to the advancement of LPWAN technologies such as LoRa [
8,
9].
LoRaWAN is an LPWAN specification protocol based on LoRa technology introduced by the LoRa Alliance [
10], which still has much room to improve its network performance. In the LoRaWAN network, the end device (ED) access adopts a pure ALOHA protocol, which adheres to a simple principle: transmit data whenever there is data to send. Although this straightforward approach simplifies LoRa network deployment, it concurrently heightens the probability of collisions during data packet transmission [
11]. As the number of connected end devices (EDs) continues to increase, large-scale data conflicts become more likely, resulting in data loss and retransmission. This, in turn, amplifies the energy consumption of EDs and substantially undermines the performance of the LoRaWAN network. Consequently, mitigating the high collision rate in the connection of large-scale EDs emerges as a primary challenge in the evolution of LoRaWAN. In response to this challenge, researchers have proposed various solutions leveraging existing LoRaWAN functionalities at the physical or MAC layer.
The configuration of LoRa physical layer transmission parameters significantly influences the network’s overall performance. Among these parameters, the spreading factor (SF) is a critical factor influencing the distance, time, and data transmission rate in LoRa networks. Consequently, the allocation strategy for SF has garnered significant attention in research. A study [
12] suggested that EDs in LoRa networks use the highest SF, ensuring comprehensive city coverage. However, this approach only considers the path loss and neglects the transmission latency when using SF12. Furthermore, several studies propose enhancements at the MAC layer of the LoRaWAN network to address resource allocation challenges within the LoRa network. In [
13], the Slotted ALOHA protocol is employed to reduce collisions. However, inadequate configuration of transmission parameters in these studies may lead to severe collision problems under high loads, ultimately compromising communication efficiency and diminishing overall network performance.
To alleviate congestion and collisions resulting from a large number of connections in the LoRaWAN network, we propose the Circular Region-Based Grouped Bit-Slot LoRa protocol. This protocol allocates transmission parameters for EDs across the entire region and completes partitioning based on different SFs. It employs the CGBS-LoRa protocol instead of the ALOHA MAC protocol to optimize resource utilization in the LoRaWAN network. By reducing collision probability, it aims to enhance the throughput of the LoRaWAN network, therefore improving overall performance.
The remaining sections of this paper are organized as follows:
Section 2 introduces related work.
Section 3 provides an overview of the LoRa physical layer and LoRaWAN protocol.
Section 4 details the CGBS-LoRa protocol, including improvements in parameter allocation and ED access algorithm.
Section 5 evaluates the proposed method through simulation and compares it with other approaches. Finally,
Section 6 concludes the paper.
2. Related Work
LoRa technology and the LoRaWAN network, a wireless technology for the IoT, have received considerable attention from the research community in recent years. Scholars have extensively explored the capabilities and performance of LoRaWAN, as detailed in [
14,
15,
16], encompassing the technology, challenges, and proposed solutions. Research [
17,
18] has examined the uplink traffic of LoRaWAN, identifying issues such as low reliability, high latency, and possibly poor throughput. We briefly describe some of the existing methods to improve the performance of LoRaWAN networks, as shown in
Table 1. Since LoRaWAN is an ALOHA-based network [
19], the probability of packet collisions increases significantly as the number of EDs increases, resulting in a decline in packet delivery rate. Consequently, it is imperative to address and improve the issue of data transmission conflicts in the LoRaWAN network.
Most existing research has primarily focused on the LoRaWAN ALOHA MAC protocol, and many different approaches have been proposed [
20,
21,
22]. In [
23], the proposal is to replace the pure ALOHA method with Slotted ALOHA to improve communication in the LoRaWAN network. In the Slotted ALOHA system, the channel time is divided into multiple slots, with each ED transmitting data packets at the beginning of a slot. The results show an overall improvement in network performance. In [
24], the authors use the Carrier Sense Multiple Access (CSMA) to enable EDs to sense the channel before transmission, reducing interference in densely deployed networks. Compared with pure ALOHA, this method provides better channel utilization. In [
25], the Distributed Queuing Random Access protocol is applied to LoRaWAN networks, which divides the terminal nodes into two queues: the collision resolution queue for collision resolution and the data transmission queue for transmitting data packets. The research shows that DQ-LoRa outperforms the pure ALOHA method in LoRaWAN concerning network throughput and managing a large number of EDs.
Additionally, research focused on enhancing the capacity of the LoRa network by adjusting transmission parameters. Analysis in [
26] shows that the reliability of the LoRaWAN network improves if EDs select appropriate spreading factors (SFs). Allocation algorithms for SFs are proposed [
27,
28]. EXPLoRa-AT [
27], for instance, divides EDs into six groups and assigns SFs to EDs through the server to balance the airtime of data packets transmitted by system EDs in each SF group. This algorithm significantly outperforms basic ADR strategies. However, in the case of a large number of connected EDs, the server cannot receive packets from EDs with the same SF due to collisions. As such, we propose enhancements to both the allocation of LoRa physical layer transmission parameters and the ALOHA-based medium access method of the LoRaWAN network.
Table 1.
Comparision of related work research.
Table 1.
Comparision of related work research.
Reference | Objective | Method Description |
---|
[19] | ALOHA | EDs send packets randomly, and under high load conditions, this randomness can result in severe collisions, therefore diminishing the success rate of data transmission. |
[23] | Slotted-LoRa | The article introduces the Slotted ALOHA scheme to improve communication in the LoRaWAN network. In the Slotted-LoRa system, each ED sends packets at the beginning of the time slot, which reduces the probability of packet collisions. |
[24] | CSMA-LoRa | The article improves network performance by reducing interference and conflicts through channel detection; it acknowledges a limitation in the CAD mechanism. The blind spot for detection may lead to channel conflicts, reducing overall channel utilization. |
[29] | ADR algorithm | The algorithm aims to optimize network performance by dynamically adjusting the data rate and transmission power of EDs. However, in large-scale networks, the requirement for independent information updates for each ED by servers may escalate collisions and reduce throughput. |
[26] | Bit Rate Ratio algorithm | The article proposes a TP and SF allocation scheme in long-range networks to enhance performance. However, the limitation of using only the default LoRa channels constrains the potential for improving network throughput through multichannel communication. |
[27] | EXPLoRa-AT | The algorithm is used for SF allocation in LoRa communication. It divides the EDs into six groups and centrally assigns the recommended SFs to each ED at the server. However, using random access may decrease the number of EDs that can reach the server as the number of EDs increases. |
3. Overview of LoRa and LoRaWAN
LoRa network consists of two different layers: LoRa and LoRaWAN. LoRa, representing the physical layer modulation, is a long-range wireless transmission technology designed by Semtech. LoRaWAN is an open standard established by the LoRa Alliance [
10], serving as the MAC layer protocol residing above the LoRa physical layer. These two parts together form the LoRa system capable of achieving LPWAN functionality. The LoRaWAN network stack is shown in
Figure 1, and it is described below.
3.1. LoRa Physical Layer
LoRa refers to the physical layer of the protocol stack, more specifically, the proprietary CSS modulation developed by Semtech [
30]. It incorporates forward error correction techniques. The characteristics of LoRa modulation are determined by several parameters, including bandwidth (BW), spreading factor (SF), coding rate (CR), and transmission power (TP) [
31]. In communication between EDs and gateway (GW), these parameters can be adjusted to modify the bit rate, interference resistance, and the range of the wireless communication link. During LoRa transmission, the bandwidth is typically configured as 125, 250, and 500 kHz [
32], and the spreading factor takes values within SF = [7, 8, 9, 10, 11, 12]. LoRa data rates are from 0.3 kbps to 50 kbps. The choice of SF represents a trade-off between coverage range and data rate. A larger SF provides a greater coverage range for EDs but at a lower data rate, whereas a smaller SF results in a higher data transmission rate but a shorter coverage range [
33,
34].
During LoRaWAN communication, a LoRa symbol comprises
chips, where the chip rate equals bandwidth (BW). Therefore, the single symbol duration
is calculated as follows [
10]:
The LoRa symbol rate
is calculated as follows:
LoRa data rate
refers to the number of bits transmitted per second and is calculated as follows:
Figure 2 shows the LoRa packet structure, including a preamble, an optional header, the data payload, and an optional cyclic redundancy check (CRC).
Time on Air (ToA) is the transmission time of a LoRa data packet from the ED to the gateway. The ToA depends on SF, BW, CR, and the transmitted data payload size. The specific equation for calculating ToA is as follows:
where
is the duration of the preamble, and
is the duration of the payload transmission.
The preamble consists of the unmodulated symbols
and 4.25 symbols marked by the end of the preamble, which serves as the first part of the LoRa packet transmission. The receiving end utilizes the preamble to identify incoming data packets. The equation for calculating the duration of preamble transmission is as follows:
Regarding the number of symbols in the LoRa data packet payload and the transmission time, they are calculated using the following equation:
where
is the payload size (1–255 bytes);
is the spreading factor (7–12);
is the cyclic redundancy check, enabled with 1 or disabled with 0, the default is
= 1;
is the header mode type, where 1 is implicit mode, and 0 is explicit mode;
is whether low data rate optimization is enabled (0 or 1);
is the coding rate, the default is 4/5. From the above equations, the ToA of the LoRa packet can be obtained.
3.2. LoRaWAN Architecture
LoRaWAN is a low-power wide-area network communication protocol running on LoRa, which defines in detail how EDs can communicate with the gateway. LoRaWAN has a data rate of up to 50 kbps and a payload of up to 250 bytes. However, the coverage of LoRa is influenced by surrounding structures and materials, and in rural areas, LoRa signals are typically transmitted over distances of up to 10 km with a battery life of about ten years.
The architecture of a LoRaWAN network adopts a typical star topology, comprising four fundamental components: End Device (ED), Gateway (GW), Network Server (NS), and Application Server (AS), as shown in
Figure 3. The work of each part is described below:
- 1.
End Device: The ED is a low-power sensor responsible for data collection. It communicates with the GW using the LoRaWAN protocol, enabling point-to-point transmission over long distances.
- 2.
Gateway: The GW is an intermediary device connecting the EDs and the Network Server. It receives data from EDs and forwards these data to the NS. In LoRa deployments, multiple GWs can be utilized, although this paper primarily considers a single gateway.
- 3.
Network Server: The NS is responsible for managing and controlling the entire network. It receives data forwarded by gateways, performs decoding and deduplication, and then sends the processed data to the appropriate AS.
- 4.
Application Server: The AS is primarily responsible for handling the data sent by the NS and processing it according to business logic
Figure 3.
LoRaWAN network architecture.
Figure 3.
LoRaWAN network architecture.
LoRaWAN supports three different types of EDs to meet the various requirements of IoT applications [
19]:
Class A: EDs randomly perform uplink transmissions and then wait to receive pending messages from the GW. This class of devices opens two short receive windows after each uplink transmission. Moreover, downlink transmissions are only allowed after uplink messages at any time. Class A has the lowest power consumption and can be used by any device.
Class B: EDs send messages on demand, allowing devices to receive signals at scheduled times. Therefore, the GW needs to periodically transmit beacons (BCN) to synchronize all devices in the network. Class B has moderate power consumption.
Class C: EDs continuously listen to the channel and keep the receive window open to receive messages from the gateway, even if they are not actively transmitting. Consequently, Class C devices have the highest power consumption.
3.3. LoRaWAN Protocol
In LoRaWAN networks, improving the scalability of the ED is possible by configuring its transmission parameters [
35]. The LoRaWAN specification introduces the Adaptive Data Rate (ADR) mechanism, automatically configuring uplink transmission parameters for optimized, reliable, and energy-efficient communication. Nevertheless, the transmission parameters of the ED can be adjusted automatically by the network server or manually configured [
29]. In the ADR mechanism, the NS completes transmission parameter configurations based on the ED’s received signal strength indication (RSSI). However, this adaptive mechanism involves significant information exchange and may increase energy consumption due to the extended adjustment process. Therefore, active allocation of transmission parameters can further improve the network performance by selecting the appropriate SF for the uplink transmission of EDs. Without the effective use of LoRa physical layer characteristics and optimization of MAC protocol, it may be difficult to achieve the best performance of the LoaWAN network. Consequently, we need to improve the ALOHA algorithm based on the rational allocation of parameters to the EDs.
4. Research Design and Methodology
4.1. Overview
As previously mentioned, the LoRa physical layer provides various transmission parameters. Rational allocation and use of these parameters can overcome most of the challenges LoRa networks face. Consequently, in this paper, we leverage these transmission parameters to their full extent and propose a protocol as an alternative to the ALOHA algorithm. The aim is to mitigate collisions, therefore optimizing the performance of the LoRa network. Our research is two-fold: (i) allocating transmission parameters that affect the scalability of LoRa networks and (ii) improving the medium access method of pure ALOHA. These aspects are described in the following sections.
4.2. Allocation of Transmission Parameters
In LoRaWAN, the allocation of network resources for large-scale deployment of EDs has received considerable attention. Among these considerations, the SF is one of the most critical parameters in the LoRaWAN network. Signals modulated with distinct SFs in LoRa are orthogonal [
24,
36], meaning non-interference when transmitted on the same channel. The SF selection range spans 7 to 12, indicating that the LoRa gateway can simultaneously receive signals from six different SFs, realizing concurrent transmissions within the LoRa network. Consequently, we propose a LoRaWAN network SF allocation method based on the distance between nodes and the gateway. By allocating different SFs to EDs, network throughput is enhanced, and information transmission is enabled. The detailed description follows below.
First, it is imperative to calculate the transmission distance relationship between devices in the LoRaWAN network. According to the theoretical equation of wireless communication:
where
is the transmission power of the transmitter,
is the sensitivity of the receiver,
is the transmitter antenna gain,
is the receiver antenna gain,
f is the carrier frequency,
d is the distance between the receiver and transmitter antennas,
c is the speed of light,
is the feeder loss of the transmitter antenna at the base station, and
is the medium loss.
Here,
and
c are constants; we carried out the conversion of the units of
f and
d to MHz and km to obtain the following expressions:
Thus, Equation (
8) can be easily converted into the following Equation (
10):
In the equation, the transmission distance d is primarily determined by , , , and f. The transmission power is limited by standard specifications. Therefore, when the medium loss and carrier frequency f are determined, the transmission distance can be altered by changing the receiver sensitivity . These parameters are calculated below.
4.2.1. Transmission Medium Loss
The wireless transmission medium loss in LoRaWAN networks is commonly influenced by factors including free space path loss, terrain, and other environmental elements. Considering these factors collectively, we employ a simplified path loss model in which loss values are estimated based on the environment, assigning a loss of 35 dBm for small towns and 40 dBm for large cities. In practical applications, we consider a total medium loss of 45 dBm.
4.2.2. Receiver Sensitivity
LoRa receiver sensitivity denotes the lowest power at which the LoRa receiver can effectively receive and decode signals. This sensitivity depends on the hardware and configuration parameters of the LoRa receiver. Among these parameters, the SF not only controls the data transfer rate but also influences the sensitivity of the LoRa receiver, with sensitivity increasing as SF increases. According to the LoRa official documentation [
37], the equation for calculating the:
where −174 is the thermal noise density, the value is mainly affected by the receiver temperature, in dBm/Hz,
is the receiver bandwidth,
is the device noise factor, influenced only by hardware performance, in dB,
is the signal-to-interference-plus-noise(SINR) Ratio required for LoRa SF demodulation, in dB [
38]. The
varies for different
values, as shown in
Table 2.
The receiver sensitivity is related to the SF and channel bandwidth. We assume a noise figure of 2 dB for the optimized LoRa gateway receiver. With a modulation bandwidth of 125 kHz and
= 12, the
is −21.5. Substituting this into Equation (
12), we obtain:
Calculate the receiver sensitivity for SF (7–12) in order, as shown in
Table 3.
4.2.3. The Relationship between SF and Distance
We set the transmit power
= 12 dBm, medium loss
= 45 dBm, and wireless frequency
f = 470 MHz through the above calculation. Based on the relationship between receiver sensitivity and SF, we choose SF = 7 corresponding to
= −130 dBm. Substituting the above parameters into Equation (
9), we obtain:
Sequentially calculate the corresponding distances for SF7 to SF12. Considering the comprehensive calculations above, we obtain the relationship between different SF values and distances, as shown in Equation (
13):
4.3. Circular Region-Based Grouped Bit-Slot LoRa (CGBS-LoRa) Protocol Description
According to the SF and transmission distance relationship, the LoRaWAN network can be partitioned into six subregions to mitigate partial collisions among EDs. Given that LoRaWAN uses the ALOHA access mechanism, the probability of collisions increases rapidly within a subregion with an increasing number of EDs. To mitigate collisions among EDs in the same circular subregion further, we propose CGBS-LoRa to replace the ALOHA protocol by time slot reservation and removing idle time slots, therefore improving the overall performance of the LoRaWAN network. The CGBS-LoRa algorithm is described below.
The CGBS-LoRa algorithm consists mainly of two phases: grouping EDs and detecting collision bit-slots. First, the GW sends a receive command to the EDs, randomly generating a bit sequence and transmitting it to the GW. The GW detects the collision bits in all received bit sequences, estimates the number of EDs, and proceeds with the grouping process. The bit sequence consists of
L bits, as shown in
Figure 4. During data transmission, the ED sets the bit-slot corresponding to the selected random number as 1, while the remaining slots are set to 0. The ED then transmits the generated bit sequence to the GW. The GW checks the received bit sequence for collision bits. A collision bit-slot indicates that at least one ED has selected the bit. If the collision bit is selected by only one ED, it is a success bit-slot, and the GW can successfully receive the data transmitted by that ED. Conversely, if more than one ED selects the collision bit, it is a collision bit-slot, and the GW cannot receive the data from the ED, requiring a retransmission in the next receive cycle. Algorithm 1 illustrates the main steps of the proposed protocol.
Algorithm 1: CGBS-LoRa Algorithm |
- 1:
Input: - 2:
: Locations of n EDs in the LoRaWAN network - 3:
: Location information of the GW - 4:
SF: Initial partition into 6 SF regions - 5:
Output: - 6:
slots: Number of slots required to complete data transmission for n EDs - 7:
procedure
PartitionSFRegions - 8:
for each ED in do - 9:
assign ED to an SF based on location - 10:
end for - 11:
end procedure - 12:
for each SF in SF do - 13:
procedure EDsReceiveAndGenerateBits - 14:
receive and generate bits - 15:
end procedure - 16:
conflict_bits ← AnalyzeReceivedBits - 17:
procedure DetectCollisionAndGroupEDs - 18:
for each ED in SF do - 19:
conflict_bits_ED ← CheckCollisionBits(ED) - 20:
estimated_EDs_ED ← EstimateEDsNumber(conflict_bits_ED) - 21:
end for - 22:
end procedure - 23:
for each ED in SF do - 24:
GroupEDs(GW, SF, estimated_EDs_ED) - 25:
end for - 26:
procedure DataTransmission - 27:
transmit data - 28:
end procedure - 29:
if ConflictOccurred then - 30:
RequestRetransmission - 31:
end if - 32:
end for - 33:
if
AllEDsTransmittedSuccessfully
then - 34:
FinishAlgorithm - 35:
end if
|
Assuming the number of EDs within the signal range received by the GW is
n, and the length of the bit sequence is
L, the probability of an ED selecting a specific bit-slot within the bit sequence is
P = 1/
L. We then consider the use of a binomial distribution to model the activities of the EDs:
Thus, the probability of a bit-slot not being selected by EDs is:
The probability of a bit-slot being selected by only one ED is:
The probability of a bit-slot being selected by multiple EDs is:
Within a receive cycle, the expected number of bit-slots selected by only one ED (successful bit-slots):
The expected number of bit-slots selected by more than one ED is (collision bit-slots):
If the GW detects collisions in Lc bit-slots, it requires Lc to receive cycles to receive data from all EDs. The data packet delivery rate of the CGBS-LoRa algorithm
is expressed as:
4.4. System Model
In this research, we consider a LoRaWAN network system model consisting of a single gateway and n end devices. Both GW and EDs are static and configured with transmission parameters. The EDs are randomly distributed in a circular region with radius R centered on GW. Based on different SFs, this region is partitioned into six circular subregions, each containing randomly distributed EDs using the same SF, as shown in
Figure 5. This strategy effectively prevents collisions among EDs located in the same gateway receiving region but in different subregions, thanks to using different SFs. Nevertheless, EDs positioned within the same subregion may still experience collisions due to using the same SF. The employment of the CGBS-LoRa algorithm further mitigates collisions among EDs.
5. Performance Evaluation
In this section, we conducted simulated evaluations of the CGBS-LoRa protocol utilizing MATLAB. The primary focus of the investigation is the process wherein n EDs transmit packets to a single gateway. To assess the performance of the LoRaWAN network, the connected EDs varied in number from 50 to 1000, randomly positioned within a 15 km radius of the gateway. Each ED generates a 10-byte packet, employing a transmission power of 12 dBm, a channel bandwidth of 125 kHz, a carrier frequency of 470 MHz, and an SF in the range of 7 to 12. The SF choice represents a trade-off between transmission distance and time, with a larger SF indicating the capability of longer-distance transmission at the expense of increased transmission latency. The completion of transmission is defined when the gateway receives all packets sent by n EDs. Simulation parameters for the LoRaWAN simulation are detailed in
Table 4. To ensure the accuracy of the results, we executed 1000 simulations and computed the average data. Equation (
9) establishes the SF-distance relationship. In the CGBS-LoRa protocol, ED devices are strategically assigned, dividing the area into zones with distinct SFs for simultaneous transmission based on their proximity to the gateway. We compare the CGBS-LoRa protocol with Slotted-LoRa [
23], DFSA-LoRa [
39], and the Adaptive Data Rate (ADR) algorithm [
29] for LoRaWAN, evaluating the packet delivery rate (PDR) and transmission delay (TD) as performance metrics.
5.1. The Probability of Successful Packet Transmission
The packet delivery rate (PDR) is a crucial metric for assessing the reliability of LoRaWAN network transmissions.
Figure 6 shows the packet delivery rate as a function of the number of EDs. As the number of EDs increases, the PDR experiences a decline across all four protocols—ADR-LoRaWAN, Slotted-LoRa, DFSA-LoRa, and CGBS-LoRa. This decline can be attributed to an increased collision probability with the growing number of EDs, consequently diminishing the reliability of data packet transmissions. When connecting some of the EDs, the ADR-LoRaWAN protocol exhibits a heightened collision probability, owing to the utilization of the ALOHA random access method, leading to a continuous decrease in PDR. However, as the number of connected EDs increases, the Adaptive Data Rate mechanism optimizes the transmission parameters, mitigating certain collision probabilities and stabilizing the PDR. In the Slotted-LoRa protocol, LoRa nodes transmit information to the gateway through time slots, theoretically reducing the probability of packet collisions. However, as the number of EDs continues to increase, a significant number of ED collisions lead to a rapid decrease in PDR. The DFSA-LoRa protocol allocates time slots to EDs, dynamically adjusting slot numbers based on collision frequencies, stabilizing the PDR at 36% with an increasing number of EDs. The CGBS-LoRa, proposed in this paper, minimizes time slot wastage by configuring ED transmission parameters and reserving slots, markedly reducing ED collisions. Notably, as the number of connected EDs increases, the PDR stabilizes at an impressive 80%.
5.2. The Transmission Delay of End Devices
Transmission Delay (TD) serves as a critical metric, representing the time required for EDs to accomplish data transmission, playing a pivotal role in assessing LoRa performance.
Figure 7 shows the transmission delay as a function of the number of EDs for the four protocols. With an increasing number of connected EDs, both ADR-LoRaWAN and Slotted-LoRa exhibit a rapid escalation in the consumption of time slots. When connecting EDs to 600, the consumed time slots surpass 5000, owing to the simplistic access method in these protocols, resulting in a heightened probability of data collisions. Consequently, a substantial number of EDs necessitate retransmission, further amplifying the transmission delay. In the DFSA-LoRa protocol, the completion of data transmission by 1000 EDs requires approximately 2600 time slots, showcasing a lower transmission delay compared to ADR-LoRaWAN and Slotted-LoRa. The CGBS-LoRa, proposed in this paper, demonstrates a significant reduction in the number of time slots needed for 1000 EDs to accomplish data transmission compared to the other three protocols. Moreover, an equivalent number of EDs completing transmission consumes only half the time slots compared to the DFSA-LoRa protocol. Consequently, among the four algorithms, CGBS-LoRa emerges with the most minimal transmission delay.
Figure 8 shows the transmission delay of EDs across the four protocols with varying spreading factor (SF) assignments, demonstrating a significant impact of different SFs on transmission latency. The Time on Air (ToA) metric denotes the time SFs takes to transmit a packet [
40], and
Table 5 provides ToA values for a 10-byte packet, 125 kHz bandwidth, and a coding rate of 4/5. Notably, the CGBS-LoRa protocol exhibits the lowest transmission latency across different SF zones. In comparison, ADR-LoRaWAN has the highest transmission latency among the four protocols. Slotted-LoRa and DFSA-LoRa, lacking the utilization of LoRa’s multichannel functionality in the physical layer, involve communication on a single channel, further increasing latency. In particular, these two protocols do not rationally allocate SFs, as all EDs start transmission with SF12 to ensure successful gateway reception within its range. However, the prolonged ToA for data transmission using SF12, coupled with increased collision probability at the same SF, leads to extended transmission delay, severely constraining LoRa network performance. We performed the same partitioning operation for both protocols for experimental comparison. Nevertheless, the transmission delay for these protocols significantly exceeds that of the CGBS-LoRa protocol across various SFs.
6. Conclusions and Future Work
This study introduces the Circular Grouped Bit-Slot LoRa (CGBS-LoRa), an innovative MAC protocol tailored for large-scale networks. CGBS-LoRa enhances network performance by mitigating collisions arising from the ALOHA access method employed by EDs and controlling the allocation of their transmission parameters. Initially, CGBS-LoRa partitions the LoRa network into six circular subregions, each characterized by different spreading factors. It then allocates transmission parameters to EDs within each subregion, facilitating concurrent transmissions in the region. Building upon this foundation, we have improved the ALOHA access method, enabling EDs within each region to communicate with the gateway using bit sequences, diminishing the probability of collisions among EDs. Simulation results show that the packet delivery rate (PDR) and transmission delay (TD) of the LoRa network are significantly improved in the case of a single LoRa gateway. Precisely, in achieving large-scale connectivity in LoRaWAN networks, CGBS-LoRa attains twice the packet delivery rate of DFSA-LoRa. Furthermore, compared to ADR-LoRaWAN, CGBS-LoRa demonstrates noteworthy advancements in PDR and latency.
During the experiments, we randomly deployed the EDs and conducted a regional partition based on SFs, resulting in an uneven distribution of EDs within the regions. The variation in SFs introduces distinct packet transmission latencies, with larger SFs leading to extended ToA. The uneven allocation of more EDs to larger SFs in this regional division prolongs the time required for data transmission and contributes to increased energy consumption. Our future work aims to dynamically adjust transmission parameters to achieve a more balanced distribution of EDs. In practical implementation, considering the ED distribution within a single gateway’s range, we will strive for a uniform zoning strategy to enhance overall performance. This study emphasizes key performance metrics like Packet Delivery Rate and Transmission Delay. Future investigations will comprehensively analyze the power consumption of LoRa systems, contributing to an enhanced understanding of power efficiency and overall system performance.