Information age optimization method under pure aloha protocol in loRa

Abstract

The present application relates to a method for optimizing the Age of Information (AoI) of data packet uplink transmission under the pure Aloha protocol in LoRa, aiming to improve the timeliness of data packets, optimize the quality of data transmission, and reduce information delay and system response time. It includes: a system model for collision and retransmission of LoRa data packet transmission process under the pure Aloha protocol is established, and the possible impact of data packet retransmission on LoRa security is analyzed. The model takes into account channel selection, spreading factor (SF) setting and possible data packet collision in LoRa network. By analyzing the transmission effect of different channel and spreading factor combinations, the collision probability of data packets in the transmission process and its impact on AoI can be calculated. In order to achieve the optimal channel and SF selection, the present application adopts the method of deep reinforcement learning, constructs a training model to dynamically select the best transmission strategy, adjusts the channel selection and spreading factor intelligently, reduces the collision events and the number of retransmissions, and thus improves the timeliness of data packets and the security of LoRa network. The present application is suitable for various LoRa Internet of Things applications, including but not limited to smart city, environmental monitoring, intelligent agriculture, intelligent logistics and other fields, especially for Internet of Things application scenarios that require high timeliness and real-time response, and has important practical value.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of Internet of Things, and particularly relates to a method for AoI optimization of data packet uplink transmission under a pure Aloha protocol in LoRa. BACKGROUND

[0002] With the rapid development of Internet of Things technology, Low Power Wide Area Network (LPWAN) has become one of the key technologies to support large-scale device connection. As a mainstream solution in LPWAN, LoRa technology has been widely used in smart cities, agricultural monitoring, environmental monitoring, logistics tracking and other fields due to its long-distance communication, low power consumption and wide application prospects. However, the communication quality and timeliness of LoRa network are still one of the key factors restricting its application, especially the delay of information in the data transmission process, which directly affects the network response speed and the real-time performance of the system.

[0003] Currently, there are few studies on AoI optimization of LoRa network at home and abroad. AoI is an important indicator to measure the freshness of data in the network, representing the time delay from data generation to reception at the receiving end. In the Internet of Things system, low AoI usually means that the network can timely reflect the changes of the external environment and support real-time decision and response. However, there is a certain time delay in the data transmission of LoRa network, especially in high-density devices and complex environments, which may lead to a longer AoI, thereby affecting the real-time performance and efficiency of the system.

[0004] Existing methods usually select channels and SFs to reduce collisions or improve signal quality through traditional greedy algorithms and traditional algorithms. However, these traditional methods often cannot fully adapt to the dynamically changing network environment. Therefore, how to intelligently adjust the channel and spreading factor of LoRa network based on the dynamic network state to optimize AoI has become a hot research topic. Deep reinforcement learning (DRL) provides a new way to solve this problem as a technology that can adaptively optimize the strategy of complex systems. By training a deep reinforcement learning model, the optimal transmission strategy can be dynamically selected in the actual network environment, not only effectively reducing the occurrence of collisions and retransmissions, but also adjusting in real time under the changing network conditions to optimize the transmission efficiency and timeliness of information. SUMMARY

[0005] A method and system for AoI optimization of data packet uplink transmission under a pure Aloha protocol in LoRa, characterized by the following steps:

[0006] S1: Through the analysis of the performance characteristics of the overall LoRa network, it is found that data packet collision is the main factor affecting information age. By using the pure Aloha protocol, data packet collision occurs frequently, and three necessary conditions for data packet collision are obtained. A model of data packet collision is constructed;

[0007] S2: Calculate the probability of two data packets being transmitted at the same rate, including the following steps:

[0008] S2-1: Calculate the rate R of the data packet b , get R b The rate is mainly determined by SF. Within the range of the gateway, if two terminals send data packets with the same SF, their rates will be the same.

[0009] S2-2: According to the traditional way, the SF of the data packet is allocated according to the distance between the terminal and the gateway. The probability and the number of data packets selecting the same SF within the range of the gateway are calculated, respectively and N SF = P SF N;

[0010] S3: According to the data packet collision model, the probability mass function of data packets selecting the same channel is

[0011] S4: According to the data packet collision model, the probability of data packet overlap is calculated.

[0012] S4-1: By calculating the air transmission time of the data packet, the air transmission time (ToA) is an important data for optimizing AoI. The ToA of the data packet is closely related to SF. The larger the SF, the larger the ToA and the stronger the anti-interference. The smaller the SF, the smaller the ToA and the weaker the anti-interference.

[0013] S4-2: Data packet overlap is the overlap of ToA between data packets. The overlap of two data packets is considered as the intersection of two Poisson processes. The probability of data packet overlap is The probability of data packet i and the rest of the data packets overlapping in the transmission process is Through the Poisson distribution process of the data packet still subject to the same parameters after collision, it is known that the retransmission probability is equal to the first collision probability, that is

[0014] S5: According to the collision model of the data packet and the calculation probability of the data packet collision, the terminal will resend the data packet when the time of sending ACK confirmation frame exceeds the time of sending ACK confirmation frame in the LoRa gateway. It is known that the data packet can be divided into first transmission and retransmission. The calculation formula of information age for first transmission and retransmission is obtained, respectively AoI=T SF and AoI=T SF +T out+P 1st *T SF ;

[0015] S6: According to the data packet collision model and the calculated data packet collision probability in S1-S5, the parameters affecting the final AoI of the system are SF and the channel, and the SAC algorithm and the TD3 algorithm of deep learning are used to allocate SF and the channel, and training is performed to optimize the system AoI.

[0016] S6-1: Map the continuous action in the SAC algorithm to the discrete, and design the action set as A={SF(i), C(i)}, wherein SF(i) is the set of SFs allocated to each terminal in the i th time slot, and C(i) is the set of channels allocated to each terminal in the i th time slot;

[0017] S6-2: Set the state set in the SAC algorithm as S={A o (i), T a (i), A(i), EX(i), R t (i)}, wherein A o (i) is the AoI set of the latest data packet sent by each terminal to the gateway in the i th time slot, T a (i) is the set of air transmission times of the data packets in the previous time slot of each terminal, A(i) is the action set allocated in the previous time slot of the data packet, EX(i) is the entire time length that the data packet in the i th time slot will experience without collision, and R t (i) represents the number of times of data packet retransmission remaining;

[0018] S6-3: According to the AoI optimization purpose of making the AoI as small as possible, the reward function takes the negative number of AoI in each round, and the reward function is designed as

[0019] The beneficial effects of the present application are:

[0020] (1) By introducing deep reinforcement learning to optimize SF selection in LoRaWAN network, the present application effectively reduces data packet collision, improves data transmission success rate and network throughput. At the same time, dynamic SF selection improves the anti-interference ability of the network, reduces the communication interruption caused by static allocation, enhances the resistance of the system to malicious attacks or faults, and improves the security of the network.

[0021] (2) The present application optimizes AoI management, reduces the delay caused by data packet retransmission, and effectively improves the real-time response ability of the system. Especially in high-density network environment, the optimization of retransmission mechanism reduces the bandwidth occupation and resource waste, at the same time, reduces the risk of data packet collision, reduces the security hidden danger caused by frequent retransmission, thereby improves the stability and security of the system.

[0022] Additional advantages, objects, and features of the application will be apparent to those skilled in the art upon examination of the following specification. It is intended that the application not be limited by any of the recited details but include all variations falling within the scope of the appended claims. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to make the objects, technical solutions and advantages of the present application clearer, the preferred embodiments of the present application will be described in detail below with reference to the accompanying drawings, in which:

[0024] Figure 1 Flow chart of the method in the detailed description of the present application

[0025] Figure 2 Schematic diagram of the distance from the gateway to the terminal in the detailed description of the present application

[0026] Figure 3 Uplink data packet transmission collision process in the detailed description of the present application

[0027] Figure 4 Data packet transmission and retransmission process in the detailed description of the present application

[0028] Figure 5 Data packet transmission of each terminal in the detailed description of the present application DETAILED DESCRIPTION

[0029] The embodiments of the present application are described below by way of specific examples. Those skilled in the art can easily understand other advantages and effects of the present application from the disclosure of the present specification. The present application can also be implemented or applied in other different embodiments, and the details in the present specification can be modified or changed in various ways based on different views and applications without departing from the spirit of the present application. The technical solutions of the present application are further introduced below in combination with the specific embodiments and the accompanying drawings.

[0030] The present application proposes a method and system for AoI optimization of data packet uplink transmission under pure Aloha protocol in LoRa. Please refer to Figure 1 The specific method flow is described as follows:

[0031] Step 1: The characteristics of the LoRaWAN are that the terminal uplink transmission data uses Aloha protocol to broadcast to the surrounding gateway. As long as the gateway within the transmission range of the terminal can receive the data packet, the embodiment obtains two conditions for data packet collision in the transmission diagram: the two data packets must be transmitted at the same rate, transmitted in the same channel, and the transmission duration overlaps each other.

[0032] It should be noted that, in order to facilitate analysis, all terminal devices in the embodiment are uniformly distributed around a gateway device, each terminal device generates data packets according to a Poisson process with a parameter λ, and there are N terminal devices, and the data packets are transmitted under the condition of a fixed bandwidth and a coding rate;

[0033] Step 2: Calculate the probability that two data packets are transmitted at the same rate, including the following steps:

[0034] Step 2-1: The same rate is that the data packets in the gateway transmit data at the same rate at the same time, and the rate R b The calculation formula is Where SF is a very important parameter in the LoRa network, different SF determines the number of original bits carried by the symbol, the value range is 7-12, the bandwidth (BW) is 125, 250 and 500 kHz, and the coding rate θ∈{1,2,3,4}, according to the formula, in order for the data packets to be transmitted at the same rate, under the condition of fixed bandwidth and coding rate, two data packets select the same SF and have the same rate;

[0035] Step 2-2: Calculate the probability of the same SF in the gateway.

[0036] It should be noted that in the traditional method, the SF of the data packet is allocated according to the distance between the terminal and the gateway, and the data packets sent by different terminals are independent of each other, referring to Figure 2 The position diagram of the gateway and the terminal in the middle, according to which, when the distance from the terminal to the gateway is 0

[0037] The probability and number of terminals selecting SF are and N SF = P SF N;

[0038] Further, according to the data packet rate R b calculated in step 2-1, the same spreading factor has the same data rate, however, this static SF allocation method will cause serious conflict between nodes that are closer. Therefore, in the present application, SF will be randomly selected through deep reinforcement learning to reduce the conflict caused by distance.

[0039] Step 3: According to the data packet collision model, calculate the probability that the data packets select the same channel, which is the frequency channel selected by the terminal in the gateway to send, it should be noted that all channel states in the embodiment are the same, and the number of main channels is M, which is {C1, C2, …, CM} M} and the probability of terminal selecting a channel is 1 / M, each terminal can only select one channel, and one channel contains multiple terminals sending data packets, so the event of terminal selecting a channel can be modeled as a multinomial distribution;

[0040] Further, c i is defined as the number of terminals selecting channel C i , where i = 1, 2,..., M, and it satisfies c1+c2+...+c M =N; the probability mass function of the multinomial distribution is Simplified summary is obtained

[0041] Step 4: According to the data packet collision model, the probability of overlapping between data packets is calculated.

[0042] Step 4-1: The air transmission time of the data packet is calculated, the air transmission time (ToA) is the time spent by the data packet in the wireless medium, the calculation formula is ToA=T preamble +T payload , wherein T preamble is the duration of the preamble, the calculation formula is T preamble =(n preamble +4.25)*T s , T payload is the payload duration, the calculation formula is T payload =n payload *T s , n preamble is the number of preamble symbols, the calculation formula is wherein T s is the symbol duration, the calculation formula is

[0043] It should be noted that PL is the payload length, SF is the spreading factor, the value range is 6-12, 6 is specially used for high-speed communication, which is not applicable in most cases, therefore, the value range of the spreading factor selected in the application is 7-12, CRC represents whether the cyclic redundancy check is enabled, CRC=1 represents that the cyclic redundancy check is enabled, otherwise, H represents whether the implicit header mode is used, H=1 represents that the implicit header mode is enabled, otherwise, DE is whether the low-rate optimization is enabled, DE=1 represents that the low-rate optimization is enabled, otherwise, CR is the coding rate, the value is {1, 2, 3, 4}, the purpose is to improve the reliability of data transmission, and additional check bits are added to allow a certain number of errors to be corrected at the receiving end.

[0044] It should be noted that the chirp spread spectrum modulation technology is used for transmission in LoRa, each symbol is transmitted through a length of 2SF The symbol time refers to the length of time that the chirp occupies in the air, which is mainly determined by the bandwidth and the spreading factor. The greater the spreading factor, the more information each symbol carries, and the longer the symbol duration, which can improve the communication distance and anti-interference, but will reduce the data transmission rate; the narrower the bandwidth, the longer the transmission time of each symbol, but it can maintain the stability of the signal at a longer distance. The bandwidth in LoRa communication can be 125 kHz, 250 kHz or 500 kHz.

[0045] The ToA is important data for optimizing the AoI. The ToA of the data packet is closely related to the SF. The greater the SF, the greater the ToA and the stronger the anti-interference. The smaller the SF, the smaller the ToA and the weaker the anti-interference.

[0046] Step 4-2: The data packets overlap, and the ToA of the data packets overlaps between the data packets. The overlap of the two data packets is regarded as the intersection of two Poisson processes.

[0047] It should be noted that in the A-type transmission mode, the behavior of different terminals sending data packets follows a Poisson distribution process with different parameters, and the data packets transmitted between terminals are independent of each other. The probability of k data packets arriving in a unit of time is where k = 0, 1, 2… and λ > 0.

[0048] Further, referring to the example diagram of data packet collision in Figure 3 , where T SF is the air transmission time of the uplink data packets transmitted with the same SF, and t is the time interval between the data packets. For the case of two data packets overlapping, it can be regarded as the intersection of two Poisson processes, i.e. the time interval between the two data packets is less than or equal to T SF , so the probability of the two uplink data packets overlapping each other is P(0, λ i T SF ) represents the probability that terminal i does not have a data packet arriving in the time interval T SF , and P(0, λ j T SF ) is similar. By substituting the probability of k data packets arriving in a unit of time into the calculation, the probability of data packet overlap is

[0049] It should be noted that under the condition of satisfying the same rate and the same channel, the number of terminals within the range of the gateway is α, and the probability of the transmission time of data packet i overlapping with the transmission time of the remaining data packets is Therefore, the probability of the first transmission collision is α is the number of terminals satisfying the same rate and channel. The data packet still obeys the same parameter Poisson distribution process after collision, and the probability of retransmission data packet collision is the same as the probability of first transmission collision, that is, P Re = P 1st , wherein P Re is the probability of collision when retransmitting.

[0050] Step 5: According to the collision model of the data packet and the calculated probability of data packet collision, the terminal retransmits the data packet when the LoRa gateway exceeds the time of sending the ACK confirmation frame. It can be known that the data packet can be divided into first transmission and retransmission.

[0051] It should be noted that if collision occurs during data packet transmission, the gateway will not send the ACK confirmation data frame to the terminal. After exceeding the terminal waiting time T out for the ACK confirmation data frame, the terminal will immediately retransmit the data frame. For details, refer to the data packet transmission and retransmission process in Figure 4 , the information age can be divided into two parts: the information age of the first transmission of the data packet and the information age of the successful retransmission of the data packet. Therefore, the calculation formula of the information age is

[0052] Further, according to the calculation formula of the information age, and referring to the data packet transmission between terminals in Figure 4 , each terminal generates data packets in Poisson distribution with λ intensity. In order to facilitate the training of the deep reinforcement learning algorithm, the time is divided according to the time slot T sl , each time slot is a step, and each time slot is allocated with a corresponding SF and channel. For example, Figure 3 , after the generation of data packet 1, terminal 1 immediately transmits the data packet, and before the data packet transmission is completed, data packet 2 is generated, and data packet 2 is discarded and data packet 1 is continued to be transmitted. It is stipulated that when two terminals select the same SF and channel and the data packets overlap, the data packets collide. For example, terminal 2 and terminal 4 in the figure select the same channel in time slot 3, and data packet 9 and data packet 5 collide. If the system stipulates that retransmission of data packets is allowed, data packet 9 and data packet 5 will start to calculate T out at the end of time slot 3, for example, if T out = T sl , data packet 9 and data packet 5 will be transmitted at the beginning of time slot 5.

[0053] Accordingly, the optimization of the information age can be decoupled into the selection of SF and channel. In this embodiment, SAC and TD algorithms are used to dynamically select SF and channel to achieve the optimization of AoI.

[0054] Step 6: According to the data packet collision model and the calculated data packet collision probability in S1-S5, the parameters affecting the final AoI of the system are SF and channel. The SAC algorithm and TD3 algorithm of deep learning are used to allocate SF and channel, and training is performed to optimize the system AoI.

[0055] Step 6-1: It should be noted that the action of deep learning is the SF and channel that the terminal needs to select each time, and the continuous action in the SAC algorithm is mapped to the discrete. Let the action set of deep reinforcement learning be A = {SF(i), C(i)}, where SF(i) is the set of SF allocated to each terminal at the ith time slot, SF(i) = {SF1(i), SF2(i)…SFN(i)}; C(i) is the set of channels allocated to each terminal at the ith time slot, C(i) = {C1(i), C2(i)…CN(i)}. n n (i)}.

[0056] Step 6-2: Let the state set be S = {A o (i), T a (i), A(i), EX(i), R t (i)}. o (i) is the AoI set of the latest data packet sent by each terminal to the gateway at the ith time slot, T a (i) is the set of air transmission times of the data packets remaining in the previous time slot of each terminal, if the data packet is not transmitted at the ith time slot, is the air transmission time remaining before the jth terminal, if the data packet before the jth terminal has been transmitted, then is 0; A(i) is the action set allocated in the previous time slot of the data packet, if the data packet in the previous time slot is still being transmitted at the ith time slot, then A j (i) is the action allocated before the jth terminal; EX(i) is the entire time slot length that the data packet that has not been transmitted at the ith time slot will experience without collision; R t (i) represents the number of times the data packet is retransmitted. In order to facilitate calculation, the state is flattened into a one-dimensional vector, and the state S is a one-dimensional array with a length of 4N.

[0057] Step 6-3: According to the AoI optimization goal of making AoI as small as possible, the reward function should be the negative value of the overall LoRa network average AoI after each step, and the reward function is designed as

[0058] ​While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. An information age optimization method under a pure Aloha protocol in LoRa, characterized in that, The method specifically comprises the following steps: S1: Through analyzing the performance characteristics of the overall LoRa network, it is found that data packet collision is the main factor affecting information age, three necessary conditions for data packet collision are obtained by using frequent data packet collision under pure Aloha protocol, and a data packet collision model is constructed; S2: Calculate the probability of two data packets being transmitted at the same rate; S3: According to the data packet collision model, the probability mass function of the data packet selecting the same channel is calculated as Wherein, assuming that all channel states are the same, the number of main channels is M, and C1, C2,..., CM represent the number of data packets in each channel. M N is the number of terminal devices within the gateway range; S4: According to the data packet collision model, calculate the probability of overlap between data packets; S5: According to the collision model of data packets and the calculated probability of data packet collision, the terminal will resend the data packet when the LoRa gateway exceeds the time of sending the ACK confirmation frame. It is known that the data packet can be divided into first transmission and retransmission, and the calculation formulas of information age for first transmission and retransmission are obtained, which are AOI=T SF and AOI=T SF +T out +P 1st *T SF , T out represents the timeout time that the terminal needs to wait, P 1st represents the probability of collision occurring in the first transmission, T SF represents the air transmission time ToA of the uplink data packet transmitted with the same spreading factor (SF). SF is defined as the ratio of the speed of the chirp after spreading to the signal rate before spreading. The range of SF is 7-12. SF affects the ToA, throughput and transmission success rate of LoRa data packet, etc. S6: According to the data packet collision model and the calculated data packet collision probability in S1-S5, the parameters affecting the final AoI of the system are SF and channel, and the SAC algorithm and TD3 algorithm of deep learning are used to allocate SF and channel to optimize the system AoI. 2.The method for information age optimization under pure Aloha protocol in LoRa according to claim 1, characterized in that, In step S1, by analyzing the performance characteristics of the overall LoRa network, three necessary conditions for data packet collision under pure Aloha protocol are obtained, and a data packet collision model is constructed accordingly. 3.The method for information age optimization under pure Aloha protocol in LoRa of claim 1, wherein, In step S2, the transmission probability between two data packets is calculated, and in step S3, the probability mass function of data packets selecting the same channel is calculated, and in step S4, the probability of overlap between data packets is calculated, and the probability of collision between data packets is obtained, and finally in step S5, the calculation formula of information age first retransmission and retransmission is obtained according to the collision model and the collision probability. 4.The method of claim 1, wherein, In step S6, according to the theoretical calculation in S5, the SAC algorithm and TD3 algorithm of deep reinforcement learning are used to allocate SF and channel, and the overall system AoI is optimized.

Images