A Real-time Packet Length Adjustment Method for Age of Information Minimization in the Internet of Things

Abstract

The present invention discloses a method for real-time adjustment of packet length for minimizing age of information in the Internet of Things, which includes a machine node and a control node. The problem of designing the state update policy is modeled as a constrained Markov decision process. By solving a linear programming problem, the optimal packet length in each system state is obtained, and the optimized scheme is stored in the machine node. In each period, the machine node uses the feedback from the control node to obtain the current channel fading information and the age of information at the control node, and adjusts the length of the transmitted packet in real time. On the premise of meeting the power consumption constraint of the machine node, the present invention uses the control node to feedback information to the machine node to help the machine node adjust the packet length in each period in real time, effectively reducing the average age of information at the control node and providing an effective means for improving the performance of the periodic state update system.

Description

Technical Field

[0001] The present invention relates to the field of status information update in the Internet of Things, and particularly to a method for adjusting packet length in a periodic status update scenario. Background Art

[0002] In recent years, the Internet of Things technology has developed rapidly, and emerging applications have emerged in an endless stream. In many Internet of Things applications, such as environmental monitoring, vehicle networking, positioning and tracking, etc., the receiving end needs to obtain the real-time status of the target. This requires the sending end to continuously collect the real-time status information of the monitored target and send it to the receiving end to update the status information of the receiving end. However, traditional indicators such as throughput and latency are difficult to effectively characterize the performance of status updates. In order to reflect the freshness of the status information obtained by the receiving end and accurately characterize the status update performance of the system, the age of information is used as an indicator for the status update system. The concept of the age of information was first proposed in (S.K. Kaul, R.D. Yates, and M. Gruteser, “Real-time status: How often should one update?” in Proc. IEEE INFOCOM, Orlando, FL, USA, 2012, pp. 2731–2735.), and it is defined as the difference between the generation time of the latest status data packet received by the receiving end and the current time.

[0003] The literature (R. Wang, Y. Gu, H. Chen, Y. Li and B. Vucetic, “On the Age of Information of Short-Packet Communications with Packet Management,” in Proc. IEEE GLOBECOM, Waikoloa, HI, USA, 2019, pp. 1-6.) points out that the packet length has a significant impact on the status update performance, and the status update performance can be effectively improved by optimizing the packet length. The literature (M. Xie, J. Gong, X. Jia and X. Ma, “Age and Energy Tradeoff for Multicast Networks With Short Packet Transmissions,” IEEE Trans. Commun., vol. 69, no. 9, pp. 6106-6119, Sept. 2021.) studies the tradeoff relationship between the age of information and energy efficiency in multicast networks, and reduces the ratio of the age of information to energy efficiency by optimizing the packet length. The literature (D. Zheng, Y. Yang, L. Wei and B. Jiao, “Decode-and-Forward Short-Packet Relaying in the Internet of Things: Timely Status Updates,” IEEE Trans. Wireless Commun., vol. 20, no. 12, pp. 8423-8437, Dec. 2021.) studies the status update with relay assistance for short-packet communications, and reduces the age of information at the receiver by optimizing the packet length.

[0004] Most of the existing methods consider non-real-time packet length adjustment methods. After setting the packet length for transmission before the status update, the sender always sends packets with the same coding length, which makes the packet length unable to be adjusted in real time according to the system status. If the sender can obtain some information in advance, such as obtaining the current channel state information and the age of information at the receiver through receiver feedback, the sender can adjust the sending strategy in real time. For example, when the channel quality is poor, increase the packet length to improve the probability of successful decoding of the status packet, and when the channel quality is good, reduce the packet length to enhance the freshness of the status packet arriving at the receiver.

[0005] In response to this, the present invention provides a method for real-time adjustment of packet length for minimizing age of information in the Internet of Things. The machine node periodically samples the status information of the target. For example, in smart agriculture, the soil sensor periodically sends the soil humidity information to the irrigator; in water quality monitoring, the water sensor in the water periodically sends the water quality information to the water conservancy center for river pollution control. Under the condition of satisfying the average transmission power consumption constraint, the machine node adjusts the length of the status data packet sent in each period according to the channel state information and the age of information fed back by the control node. The present invention models the problem of real-time adjustment of packet length as a constrained Markov decision process, and transforms this problem into a linear programming problem through variable substitution, and gives the optimal packet length adjustment scheme based on this. Based on the obtained packet length adjustment scheme, in the actual state update process, the machine node can find and select the optimal data packet length according to the current channel state and the age of information at the control node to send the status data packet. This method for real-time adjustment of packet length can be widely applied to energy-constrained point-to-point state update scenarios, but is not limited to the above-listed scope. Summary of the Invention

[0006] The object of the present invention is to provide a method for real-time adjustment of packet length for minimizing age of information in the Internet of Things, which makes full use of the age of information and channel state information at the control node to adjust the length of the data packet sent by the machine node, and effectively reduces the age of information at the control node and improves the state update performance of the system on the premise of satisfying the average transmission power consumption of the machine node.

[0007] A method for real-time adjustment of packet length for minimizing age of information in the Internet of Things includes a machine node and a control node, where both the machine node and the control node are equipped with single antennas; the machine node periodically samples the target status information every L time slots, encapsulates it into a data packet and sends it to the control node to update the status information at the control node; within each period, the control node determines the length of the data packet sent by the machine node according to the current channel fading and the age of information at the control node; it includes the following steps:

[0008] Step 1: Design of the state update scheme: For each possible system state, optimize the corresponding data packet length to be sent, and store the optimized scheme in the machine node;

[0009] Step 2: Acquisition of channel state information and age of information: Before the start of each period, the control node feeds back the current channel fading to the machine node through the feedback channel; after the control node successfully receives the data packet, it feeds back a successful reception signal to the machine node, otherwise it feeds back a failed reception signal, and the machine node calculates the age of information of the control node in each time slot according to this feedback signal;

[0010] Step 3: At the beginning of each cycle, the machine node samples the status information. The machine node determines the corresponding status of the system according to the channel fading information and the age of information fed back by the control node, searches for the optimal packet length in this status, and encodes the collected status information into a data packet based on this, and sends it to the control node;

[0011] Step 4: The control node updates the status information: If the control node successfully receives a data packet in a certain time slot, the control node replaces the original data packet and analyzes the current status of the target using the newly obtained data packet; then repeat Steps 2, 3, and 4; the control node continuously feeds back information to the machine node, and the machine node determines the length of the data packet sent in each cycle according to the feedback information and sends a status data packet to the control node.

[0012] Compared with the existing status update method, the present invention has the following advantages and obvious effects:

[0013] In the method for real-time adjustment of packet length for minimizing the age of information in the Internet of Things of the present invention, at the packet length design stage, this method models the status update strategy design problem as a constrained Markov decision process, obtains the optimal data packet length in each system status by solving a linear programming problem, and stores the optimized scheme in the machine node. In the data transmission stage, the control node feeds back the current channel status information and the age of information to the machine node. The machine node obtains the current system status according to the feedback information, searches for and selects the optimal status data packet coding length. Finally, the machine node sends a status data packet to update the status information at the control node.

[0014] The present invention makes full use of the age of information and channel status information at the control node to adjust the length of the data packet sent by the machine node. On the premise of meeting the average transmission power consumption of the machine node, it effectively reduces the age of information at the control node and improves the status update performance of the system. Description of the Drawings

[0015] Figure 1 is the system model diagram of the present invention.

[0016] Figure 2 is the schematic diagram of the change of the age of information of the present invention.

[0017] Figure 3 is the structural diagram of the optimal packet length adjustment scheme of the machine node in the present invention.

[0018] Figure 4 is the relationship diagram of the average age of information at the control node changing with the transmission power consumption constraint.

[0019] Figure 5 is the relationship diagram of the average age of information at the control node changing with the amount of transmitted information. Detailed implementation manners

[0020] The following further elaborates on this invention patent in conjunction with the attached drawings of the specification.

[0021] As Figure 1 shown, the system of this invention includes a machine node and a control node. First, optimize the packet length for all possible states of the system, and store the optimized packet length in the machine node. During the state update process, the control node feeds back the channel state information and the age of information to the machine node. In each cycle, the machine node, based on the information obtained from the feedback, searches for the optimal packet length for the corresponding state, and encodes the state information into a packet and sends it to the control node to update the state information at the control node. Figure 2 shows the schematic change of the age of information on the control node side over a period of time. In a certain time slot, when the control node successfully receives a packet, the age of information is reduced to the age of the received packet; otherwise, the age of information is incremented by 1.

[0022] The method for real-time adjustment of packet length for minimizing the age of information in the Internet of Things of this invention specifically includes the following steps:

[0023] Step 1: Design of the state update scheme

[0024] After discretizing the wireless channel, the set of small-scale channel fading coefficients is where K is the number of channel fades after discretization. In one cycle, the probability of the small-scale channel fade h k appearing is ω k , The transmission power of the machine node is P, the distance between the machine node and the control node is d, and in a certain cycle, the small-scale channel fading coefficient is h k , then the signal-to-noise ratio when the control node decodes the state packet sent by the machine node is where α is the path fading factor, χ0 is the shadow fading, and σ 2 is the power of additive white Gaussian noise.

[0025] The channel is assumed to be a block fading channel, that is, the channel fade remains unchanged within one cycle and randomly changes between different cycles. Considering that the packet length sent by the machine node is short, for this, this invention uses the short packet theory to characterize the transmission performance of the state packet. When the channel is in the kth fading state, the signal-to-noise ratio at the control node is γ k , when the machine node sends a packet with a length of l, the decoding error packet rate of the control node is

[0026]

[0027] Among them, \(n_0\) is the block length of one time slot, \(D\) is the amount of state information sent by the machine node, and \(Q\) is the right-tail function of the standard normal distribution.

[0028] In the present invention, we use the age of information to characterize the freshness of the state information at the control node. Only when the machine node sends a state data packet and it is successfully received by the control node, the age of information at the control node becomes the age of the received state data packet, and the age of a data packet with length \(l\) sent by the machine node when it is successfully received by the control node is \(l\); otherwise, the age of information at the control node is incremented by 1. The relationship of the change in the age of information at the control node can be obtained as

[0029]

[0030] where \(\Delta(q)\) and \(\Delta(q + 1)\) are the age of information at the control node in the \(q\)-th and \((q + 1)\)-th time slots respectively.

[0031] When the machine node adopts the transmission policy \(\pi\), the average age of information at the control node can be expressed as

[0032]

[0033] where \(\gamma\) is the number of time slots passed for the state update. The transmission policy \(\pi\) is the mapping between the system state and the actions taken by the machine node, that is, the length of the data packet that the machine node should send under different channel fading information and age of information.

[0034] In the \(t\)-th period, when the machine node sends a data packet with length \(a\) t the transmission energy consumption of the machine node is \(a\) t \(P\), then the average transmission power consumption of the machine node in each time slot can be expressed as

[0035]

[0036] where \(T\) is the number of periods passed for the state update process, and \(L\) is the length of the sampling period.

[0037] The present invention obtains the optimal data packet length adjustment scheme by optimizing the following optimization problem.

[0038]

[0039]

[0040] where \(P\) c is the constraint on the statistical average transmission power consumption of the machine node.

[0041] The present invention models the real-time optimization problem of the state data packet length as a Markov decision process with constraints, and its state space, action space, state transition probability, and cost function are respectively:

[0042] State space: The machine node adjusts the packet length according to the current instantaneous channel state information and the age-of-information at the control node. The age-of-information at the control node in the second time slot of each cycle is not a continuous integer, and its value set is an arithmetic progression {1, L + 1, 2L + 1, …}. Define Δ t as the instantaneous age-of-information at the control node in the second time slot of the (t - 1)-th cycle, which can be expressed as Δ t = δ t L + 1, where δ t is the age-of-information aging degree parameter at the control node in the second time slot of the (t - 1)-th cycle, and L represents the number of time slots in a cycle. Among them, the system state s t in the t-th cycle is s t = {δ t , r t , k t}, where δ t ∈ {1, 2, …, δ m}, r t ∈ {1, 2, …, L + 1}, k t ∈ {1, 2, …, K}. Among them, to ensure that the state space is countable, δ m is the upper bound of the set value of δ t . The state information at the control node in the (t - 1)-th cycle is updated after r t time slots, that is, the age-of-information at the control node in the (t - 1)-th cycle decreases at the end of the r t -th time slot. r t = L + 1 means that the information at the control node is not updated in the (t - 1)-th cycle, and k t represents the channel fading state that the channel is in the t-th cycle.

[0043] Action space: In each cycle, the node selects the length of the packet to be sent in units of time slots. In the t-th cycle, the length of the packet sent by the node is a t ∈ {0, 1, 2, …, L}, where a t = 0 means that the node does not send a packet. The length of the packet sent in the t-th cycle is a t , and the unit is time slots. Since the cycle length is L time slots, the maximum possible value of a t is L.

[0044] Transition probability: The transition probability Pr(s t+1 |s t , a t ) represents that the system state in the t-th cycle is s t = (δ t , r t, k t ) When the machine node sends a data packet of length a t , the system state in the (t + 1)-th period becomes s t+1 with a probability, which can be specifically expressed as

[0045] When a t = 0,

[0046]

[0047] When a t = 1,

[0048]

[0049]

[0050] When 1 < a t ≤ L,

[0051]

[0052]

[0053] Among them, is the probability that the channel fading state in the (t + 1)-th period is k t+1 When r t ≤ L, δ t+1 = 1. When r t = L + 1, δ t+1 = δ t + 1, and the remaining state transition probabilities are 0.

[0054] Cost function: In the present invention, the transmission energy consumption of a node in one period is used as the cost function. In the t-th period, when the system state is s t , the machine node sends a status data packet of length a t , and its transmission energy consumption is a t P.

[0055] Based on the state space of the Markov decision process with constraints modeled above, the optimization problem in Equation (5) can be expressed as

[0056]

[0057] Based on the above Markov decision process with constraints, the present invention designs a probabilistic state update scheme. Since the length of the status data packet sent in one period is only related to the system state of that period and has nothing to do with the number of periods, the subscript t representing the number of periods can be omitted. Let Denote the probability that a machine node sends a data packet of length \(l\) when the system state is \(s = (\delta,r,k)\). Let \(\Pr(\delta',r',k'|\delta,r,k)\) denote the probability of transitioning from state \(s = (\delta,r,k)\) to state \(s' = (\delta',r',k')\). According to equations (6)-(10), the following state transition probability expressions are obtained as

[0058] When \(\delta' = 0\),

[0059]

[0060] When \(\delta' = 1\) and \(r\leq L\),

[0061]

[0062]

[0063] When \(1\lt\delta'\lt\delta\) m , \(r = L + 1\),

[0064]

[0065]

[0066] When \(\delta'=\delta\) m , \(r = L + 1\),

[0067]

[0068]

[0069] Let \(\mu\) δ,r,k denote the probability that the system is in state \(s = (\delta,r,k)\) after stabilization. Introduce the steady-state vector where is the transpose symbol. Further introduce the state transition matrix \(Q\) such that \(Q\mu=\mu\). The state transition matrix can be expressed as

[0070]

[0071] where the and \(0\) in equation (19) are both matrices of dimension \((L + 1)K\times(L + 1)K\). To facilitate the explanation of the elements in \(Q\) i,j , introduce the auxiliary vector \(\varPhi=\{(1,1),(2,1),\cdots,(L + 1,1),(1,2),\cdots,(L + 1,K)\}\). Then the element in the \(x\)-th row and \(y\)-th column of the matrix \(Q\) i,j is

[0072] Q i,j(x,y) = Pr(j, Φ(y)|i, Φ(x)). (20)

[0073] In the proposed probabilistic scheme, by optimizing to obtain the final optimized scheme. Since the distribution probability μ δ,r,k is related to the designed scheme, the present invention jointly optimizes and μ δ,r,k to design a packet adjustment scheme. Meanwhile, introducing an auxiliary vector the optimization problem in Equation (11) is transformed into the following linear programming problem.

[0074]

[0075]

[0076]

[0077]

[0078] 0 ≤ μ δ,k,r ≤ 1 (21e)

[0079]

[0080] Qμ = μ (21g)

[0081] By solving the above optimization problem, the optimal values η * and μ * of η and μ can be obtained, and then the optimal packet length adjustment probability f * can be obtained. The specific process of the algorithm is as follows.

[0082]

[0083]

[0084] The machine node stores the obtained optimal packet length adjustment scheme for real-time search of the optimal packet length in each state during the state update process.

[0085] Step 2: Obtaining channel state information and age of information

[0086] Before the start of each cycle, the control node estimates the current channel fading based on the pilot sent by the machine node and feeds back the estimated channel fading to the machine node. The machine node looks up the corresponding discretized channel fading as the current channel fading state. In addition, the control node feeds back the decoding result to the machine node in each time slot. If the receiving end successfully decodes a status data packet, the control node feeds back a signal indicating successful reception; otherwise, it feeds back a signal indicating failed reception. The machine node calculates the age of information of the control node in each time slot based on this feedback signal.

[0087] Step 3: The machine node sends a status data packet

[0088] At the beginning of each cycle, the machine node samples the status information. The machine node determines the corresponding state of the system according to the channel fading information and the age of information fed back by the control node, looks up the optimal packet length in this state, and encodes the collected status information into a data packet based on this, and sends it to the control node.

[0089] Step 4: The control node updates the status information

[0090] Since the latest data packet can better reflect the current state of the monitored target, when the control node successfully receives a new data packet, it will replace the old data packet and analyze the current state of the target using the newly obtained data packet; then repeat steps 2, 3, and 4. The control node continuously feeds back information to the machine node. The machine node determines the length of the data packet to be sent in each cycle according to the feedback information and sends a status data packet to the control node. This continues until the status update task ends.

[0091] The present invention provides a method for real-time adjustment of packet length for minimizing the age of information in the Internet of Things. The machine node periodically samples the status information of the target. For example, in smart agriculture, the soil sensor periodically sends soil humidity information to the irrigator; in water quality monitoring, the water sensor in the water periodically sends water quality information to the water conservancy center for river pollution control. Under the condition of satisfying the average transmission power consumption constraint, the machine node adjusts the length of the status data packet sent in each cycle in real time according to the channel state information and the age of information fed back by the control node.

[0092] The present invention models the problem of real-time adjustment of packet length as a constrained Markov decision process, and transforms this problem into a linear programming problem through variable substitution, and gives the optimal packet length adjustment scheme based on this. Based on the obtained packet length adjustment scheme, in the actual status update process, the machine node can look up and select the optimal packet length according to the current channel state and the age of information at the control node to send the status data packet. The method for real-time adjustment of packet length of the present invention can be widely applied to energy-constrained point-to-point status update scenarios, but is not limited to the above-listed scope.

[0093] A method for real-time adjustment of packet length for minimizing age of information in the Internet of Things according to the present invention. This method makes full use of channel state information and the age of information at the control node, optimizes the encoding length of status data packets in each period, and improves the status update performance in an energy-constrained scenario. This method models the status update strategy design problem as a constrained Markov decision process. Further, through variable substitution, the optimization problem of the status data packet length is transformed into a linear programming problem, and the optimal packet length adjustment scheme is obtained. Based on this scheme, the machine node can adjust the length of the data packet sent in each period in real time according to the channel state and the age of information fed back by the control node during the actual status update process, thereby effectively improving the freshness of information at the control node and reducing the age of information at the control node.

[0094] Embodiment:

[0095] Simulation parameter settings: The value set of the discretized small-scale channel fading coefficient is {0.4, 0.5, 0.6, 0.7}. For the convenience of simulation design, the probability of occurrence of each small-scale channel fading is the same in this embodiment. It is worth mentioning that the method proposed in the present invention is also applicable to the case where the probabilities of occurrence of each small-scale fading are different. The distance between the machine node and the control node is d = 150m, the path fading factor is α = 3.8, the shadow fading is χ0 = -50dB, the transmission power of the machine node is P = 0.2W, and the average transmission power consumption constraint for each time slot is P c = 0.1W, the amount of information transmitted by the machine node is D = 200 nats, the block length for each time slot is n0 = 5 channel uses, and one period contains 30 time slots. The channel bandwidth is 180kHz, and the power spectral density of white noise is -174dBm / Hz.

[0096] Figure 3 Shows the structure of the optimal packet length adjustment scheme obtained by using the real-time packet length adjustment method of the present invention. Among them, the circles, squares, left triangles, diamonds, right triangles, and triangles represent that the machine node sends data packets with lengths of 0, 18, 19, 20, 21, and 23 time slots respectively. The small-scale fading of channel state 1 to channel state 4 gradually decreases, that is, the channel quality gradually improves. From Figure 3It can be seen that as the channel quality gradually improves, the length of the status data packets sent by the machine nodes gradually decreases. This is because after the channel quality improves, even if the machine nodes reduce the data packet length, the decoding performance is still good. At this time, the machine nodes can effectively improve the freshness of the data packets reaching the control node by reducing the data packet length, thereby improving the status update performance. In addition, in many states, the machine nodes do not send data packets. On the one hand, when the channel quality is poor, even if the machine nodes consume energy to send data packets, it is very likely that they cannot be successfully received by the control node. At this time, the machine nodes choose to remain silent to save energy. On the other hand, the probability of occurrence of many states in the actual status update process is extremely low. At this time, the machine nodes will not optimize the data packet length for these states in advance.

[0097] Figure 4 It shows the relationship between the average age of information and the average power consumption constraint under two methods. The proposed method is the method for real-time adjustment of packet length of the present invention, and the fixed packet length method is that each cycle the machine nodes send the same data packet length on the premise of meeting the power consumption constraint. It can be found from the figure that the method for real-time adjustment of packet length of the present invention is significantly better than the fixed packet length method. Especially when the average power consumption constraint value is small, that is, the constraint is strong, it is difficult for the fixed method to find a feasible data packet length to meet the power consumption constraint. The proposed method effectively saves energy by not sending status data packets when the channel quality is poor, and selects to send status data packets when the channel quality is good, effectively reducing the average age of information of the control node.

[0098] Figure 5 It shows the relationship between the average age of information and the amount of transmitted information under two methods. From Figure 5 It can be seen that the average age of information increases as the amount of transmitted information increases. This is because as the amount of transmitted information increases, the decoding error packet rate of the control node increases, resulting in a deterioration of the status update performance. However, the increase in the amount of transmitted information also enables the control node to obtain more status information about the monitored target. Therefore, in practical applications, it is necessary to reasonably select the amount of transmitted information according to the specific requirements of the specific application.

[0099] The description of the above embodiments is relatively specific and detailed, but it only represents a feasible implementation manner of the present invention, and does not limit the scope of the patent of the present invention. It should be noted that scientific researchers and engineering personnel in this field can add several deformations or improvements on the basis of this embodiment within the framework of the present invention, but these are all within the protection scope of the patent of the present invention, and the protection scope of the patent of the present invention shall be subject to the appended claims.

Claims

1. A real-time packet length adjustment method for minimizing the age of information in the Internet of Things, characterized in that, It includes a machine node and a control node, both of which are equipped with a single antenna. The machine node samples the target state information every L time slots periodically, encapsulates it into a data packet and sends it to the control node to update the state information at the control node. Within each period, the control node determines the length of the data packet sent by the machine node according to the current channel fading and the age of information at the control node. The steps are as follows: Step 1: Design of the state update scheme: For each possible system state, optimize the corresponding length of the data packet to be sent, and store the optimized scheme in the machine node. Step 2: Acquisition of channel state information and age of information: Before the start of each period, the control node feeds back the current channel fading to the machine node through the feedback channel. After the control node successfully receives a data packet, it feeds back a successful reception signal to the machine node, otherwise it feeds back a failed reception signal. The machine node calculates the age of information of the control node for each time slot based on this feedback signal. Step 3: At the beginning of each period, the machine node samples the state information. The machine node determines the corresponding state of the system according to the channel fading information and the age of information fed back by the control node, searches for the optimal data packet length in this state, and encodes the sampled state information into a data packet based on this, and sends it to the control node. Step 4: The control node updates the state information: If the control node successfully receives a data packet in a certain time slot, the control node replaces the original data packet and analyzes the current state of the target using the newly obtained data packet. Then repeat Steps 2, 3, and 4; the control node continuously feeds back information to the machine node. The machine node determines the length of the data packet sent in each period according to the feedback information and sends a state data packet to the control node. Specifically: The optimization problem of data packet length adjustment is modeled as a constrained Markov decision process. The constrained Markov decision process includes: state space, action space, transition probability, and cost function. The state space in the constrained Markov decision process is as follows: The machine node adjusts the packet length according to the current instantaneous channel state information and the age of information at the control node. The age of information at the control node in the second time slot of each period is not a continuous integer, and its value set is an arithmetic sequence {1, L + 1, 2L + 1, …}. The instantaneous age of information Δ t at the control node in the second time slot of the (t - 1)-th period is denoted as Δ t = δ t L + 1, where δ t is the aging degree parameter of the age of information at the control node in the second time slot of the (t - 1)-th period, and L represents the number of time slots in a period; The system state s t in the t-th period is s t = {δ t , r t , k t}, where δ t ∈ {1, 2, …, δ m}, r t ∈ {1, 2, …, L + 1}, k t ∈ {1, 2, …, K}. Here, to ensure that the state space is countable, δ m is the upper bound of the set value of δ t set, r t represents that the state information at the control node in the (t - 1)-th period is updated after r t time slots, that is, the age of information at the control node in the (t - 1)-th period decreases at the end of the r t -th time slot. r t = L + 1 represents that the information at the control node is not updated in the (t - 1)-th period, and k t represents the channel fading state that the channel is in the t-th period, and K is the number of discretized channel fading states; The action space in the constrained Markov decision process is as follows: In each period, the node selects the length of the data packet to be sent in time slots. In the t-th period, the length of the data packet sent by the machine node is a t ∈ {0, 1, 2, …, L}; The transition probability in the constrained Markov decision process. The specific process is as follows: The transition probability Pr(s t+1 |s t ,a t ) represents that when the system state at the t-th period is s t =(δ t ,r t ,k t ), the machine node sends a data packet of length a t , and the probability that the system state at the (t + 1)-th period becomes s t+1 =(δ t+1 ,r t+1 ,k t+1 ). Specifically, when a t = 0, When a t = 1, When 1 < a t ≤ L wherein, and are the packet error rates when the machine node sends data packets with lengths of 1 and a t time slots respectively when the channel fading state is k t , is the probability that the channel fading state in the (t + 1)-th period is k t+1 . When r t ≤ L, δ t+1 = 1. When r t = L + 1, δ t+1 = δ t + 1, and the remaining state transition probabilities are 0; The cost function in the constrained Markov decision process is as follows: The transmission energy consumption of the machine node within one period is used as the cost function. In the t-th period, when the system state is s t at that time, the machine node sends a status data packet with a length of a t , and its transmission energy consumption is a t P, where P is the transmission power of the machine node.

2. The real-time packet length adjustment method for minimizing the age of information in the Internet of Things according to claim 1, characterized in that, To avoid affecting the transmission of data packets in the next period, the length of the data packet sent by the node within a period cannot exceed the period length. The channel is assumed to be a block fading channel, that is, the channel fading remains unchanged within a period and randomly changes between different periods. The short packet transmission theory is used to characterize the transmission performance of the state data packet. For a certain period, if the channel is in the k-th channel fading state and the machine node sends a data packet with length l, its packet error rate is expressed as where n0 is the block length of a time slot, γ k is the signal-to-noise ratio at the control node when the channel is in the k-th fading state, D is the amount of state information sent by the machine node, and Q is the right-tail function of the standard normal distribution.

3. The real-time packet length adjustment method for minimizing the age of information in the Internet of Things according to claim 2, characterized in that, When the control node successfully receives a data packet, the information at the control node is updated, and its age of information becomes the age of the received state data packet. Otherwise, the age of information is incremented by 1.

4. The real-time packet length adjustment method for minimizing the age of information in the Internet of Things according to claim 3, characterized in that, To limit the average transmission power consumption of the machine node, when the transmission policy π is adopted, the average transmission power consumption of the machine node is where the transmission policy π is a mapping between the system state and the actions taken by the machine node, that is, the length of the data packet that the machine node should send under different channel fading information and age of information, T is the number of cycles elapsed for state update, is the expectation of the total energy consumption of the machine node in T cycles when the transmission policy π is adopted. The average power consumption of the machine node needs to satisfy P c is the statistical average transmission power consumption constraint of the machine node.

5. The real-time packet length adjustment method for minimizing the age of information in the Internet of Things according to claim 4, characterized in that, When the system is in the state s = (δ, r, k), the probability that the machine node sends a data packet of length l is and transforms the packet length design problem into a linear programming problem to optimize all states in the state space to obtain the optimal packet length adjustment scheme.

Images