A Pilot-Assisted State Update Method under Short Packet Communication Conditions

Abstract

The present invention discloses a pilot-assisted status update method under short data packet communication conditions. The system includes a machine node and a control node, and both nodes are equipped with single antennas. To help the control node master the current status of the machine node, the machine node needs to send short data packets carrying status information to the control node. Due to the dynamic change of the channel, the machine node needs to periodically send pilot signals to the control node to assist the control node in obtaining the current channel status information. On the premise of meeting the system power consumption constraint, the present invention utilizes the feedback information of the control node to effectively reduce the age of information at the control node, providing an effective means to improve the freshness of the status information obtained at the control node and improve the system status update performance.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication state update, and particularly to a pilot-assisted state update method under short data packet communication conditions. Background Art

[0002] With the development of the Internet of Things, more and more real-time applications have emerged in our lives. A key issue is how to ensure that the destination can timely obtain the status information of the target to complete tasks such as monitoring and control. The machine node collects the status information of the target and sends it to the destination to replace the outdated status information at the destination, so that the destination can master the latest status of the target. However, some traditional performance metrics such as throughput and packet error rate are difficult to effectively reflect the performance of state update.

[0003] For this purpose, the present invention uses the age of information proposed by S.K. Kaul et al. (S.K. Kaul, R.D. Yates, and M. Gruteser, “Real-time status: How often should one update?” in Proc. IEEE INFOCOM, 2012, pp.2731–2735.) as a performance metric. The paper [Y. Gu, H. Chen, Y. Zhou, et al., “Timely Status Update in Internet of Things Monitoring Systems: An Age-Energy Tradeoff,” IEEE Internet of Things J., vol. 6, no. 3, pp. 5324-5335, Jun. 2019.] studied the status update performance under the truncated automatic repeat request protocol and analyzed the tradeoff relationship between the age of information and energy consumption. The paper [H. Zhang, Y. Kang, L. Song, et al., "Age of Information Minimization for Grant-Free Non-Orthogonal Massive Access Using Mean-Field Games," IEEE Trans. Commun., vol. 69, no. 11, pp. 7806-7820, Nov. 2021.] studied the status update performance in a multi-channel uplink grant-free massive access system and used mean-field evolutionary games to minimize the age of information of the system. Most of the existing studies on status update systems are based on the traditional Shannon formula. However, in the Internet of Things, the status data packets sent by machine nodes are often short. At this time, the Shannon formula is difficult to accurately characterize the transmission performance of status data packets, which limits the application of most existing status update methods in the Internet of Things scenario. In addition, most of the existing status update methods are offline. When nodes perform status updates, they cannot obtain the real-time information of the system in advance, such as channel state information, the age of information of the destination, etc. This makes it impossible for machine nodes to adjust their own status update behaviors in real time, which may lead to a waste of communication resources and cannot effectively reduce the age of information of the destination.

[0004] In response to this, the present invention provides a real-time online status update method. The machine node adjusts its own status update strategy in real time according to information such as the channel state information, age of information, and position of the next time slot in a coherence time fed back by the control node. After completing channel estimation, considering that the machine node has limited energy and cannot always send status data packets for status update, the machine node can send status data packets when the age of information at the destination is relatively high and the channel quality is good, and remain silent when the age of information is relatively low and the channel quality is poor, so as to effectively improve the status update performance of the system under the premise of meeting the power consumption constraint. This status update method is applicable to the situation where the machine node can independently send pilots, and can be widely applied to the energy-constrained point-to-point status update scenario, but is not limited to the above-listed scope. Summary of the Invention

[0005] The object of the present invention is to provide a pilot-assisted status update method under short data packet communication conditions. This method makes full use of information such as the age of information at the control node to effectively improve the freshness of information at the control node and improve the status update performance in the energy-constrained scenario.

[0006] A pilot-assisted status update method under short data packet communication conditions includes the following steps:

[0007] Step 1: Determine the coherence time: The machine node determines the coherence time of the system according to parameters such as its relative moving speed with the control node and the radio signal wavelength.

[0008] Step 2: Channel estimation: The machine node calculates the number of time slots passed within a coherence time, so as to send a pilot to the control node at the first time slot of each coherence time. The control node estimates the channel state information of the current coherence time according to the pilot.

[0009] Step 3: Perform status update: After the machine node sends the pilot, within the same coherence time, it sends a status data packet to the control node or remains silent. If the machine node sends a status data packet and it is successfully received by the control node, the information of the control node is updated and the age of information is reduced to 1; otherwise, the age of information is increased by 1.

[0010] Step 4: Control node feedback information: At the first time slot of a coherence time, the machine node sends a pilot, and the control node feeds back the obtained channel state information and the current age of information to the machine node. In the subsequent time slots of the same coherence time, since the channel fading remains unchanged within the same coherence time, at this time, the control node only feeds back the current age of information without the need to feed back the channel state information.

[0011] Step 5: Design of the status update strategy: Before the start of the next time slot, the machine node determines the transmission strategy for the next time slot based on information such as the current channel state information, the current age of information of the control node, and the position of the next time slot within the coherence time: 1. Transmit a pilot, 2. Transmit a status data packet, 3. Remain silent; thereafter, the system repeats Step 4 and Step 5 until the status update task ends.

[0012] The present invention models the problem of designing the status update strategy as a constrained Markov decision problem. By cleverly designing the state transition probability, it ensures that the machine node transmits a pilot in the first time slot of each segment of coherence time, guaranteeing that the system can obtain the channel state information within the coherence time, and adopts the method of recursive correlation values to obtain the optimal status update method. The machine node uses information such as the age of information at the control node, the channel state information, and the position of the next time slot within the coherence time to adjust the status update behavior in real time. When the age of information at the control node is relatively high and the channel state is good, it transmits the status data packet, effectively improving the status update performance on the premise of meeting the power consumption constraint.

[0013] The present invention makes full use of information such as the age of information at the control node, the channel state information, and the position of the next time slot within the coherence time to adjust the status update behavior of the machine node. On the premise of ensuring that the power consumption constraint of the machine node is met, it effectively reduces the age of information at the control node and improves the status update performance of the system. Description of the Drawings

[0014] Figure 1 is the system model diagram of the present invention, where there is one machine node and one control node, and all nodes are equipped with single antennas.

[0015] Figure 2 is the schematic diagram of the change in the age of information of the present invention.

[0016] Figure 3 is the update strategy of the machine node in the present invention under different ages of information and channel fading when the fixed time slot position is the 10th time slot within the coherence time.

[0017] Figure 4 is the update strategy of the machine node in the present invention under different time slot positions and channel fading when the fixed age of information of the control node is 20.

[0018] Figure 5 is the schematic diagram of the performance comparison between the present invention and the random status update strategy. Detailed Embodiment

[0019] The following further elaborates on the present invention patent in conjunction with the drawings of the specification.

[0020] As Figure 1As shown in the figure, the system of the present invention has a machine node and a control node, and all nodes are equipped with single antennas. The machine node uses information such as the age of information on the control node side, the current channel state information, and the position of the next time slot within the coherence time fed back by the control node, and adopts a method of recursive correlation values to obtain the optimal state update strategy, and adjusts the transmission behavior in real time during the state update process: 1. Transmit pilots; 2. Transmit status data packets; 3. Keep silent. When the node transmits a status data packet and it is successfully received by the control node, the age of information at the control node drops to 1, otherwise the age of information increases by 1. Figure 2 shows the change of the age of information on the control node side over a period of time.

[0021] A pilot-assisted state update method under short data packet communication conditions of the present invention, and the specific implementation process of the transmission method is as follows:

[0022] Step 1: Determine the coherence time

[0023] The machine node determines the coherence time of the system according to parameters such as the relative moving speed between it and the control node and the radio signal wavelength. The coherence time can be expressed as where λ is the signal wavelength and v is the relative moving speed between the machine node and the control node. Assume that the length of a time slot is n channel uses, and its size can be expressed as n = Bt, where B is the system bandwidth and t is the duration of a time slot. The units of bandwidth and time are s and Hz respectively. Therefore, the number of time slots included in a period of coherence time can be expressed as

[0024] Step 2: Channel estimation

[0025] Since the channel fading changes within different coherence times, for this, the machine node sends pilots to the control node in the first time slot of each period of coherence time. Assume that the transmission power of the pilots is P 1 , and the control node estimates the channel fading of the current coherence time according to the pilots and feeds back the channel state information to the machine node.

[0026] Step 3: Perform state update

[0027] After the machine node sends the pilots, within the same coherence time, it sends a status data packet to the control node or keeps silent. If the machine node sends a status data packet and it is successfully received by the control node, the information at the control node is updated and the age of information drops to 1, otherwise the age of information increases by 1.

[0028] After the wireless channel is discretized, the set of channel fades obtained is The probability of the channel fade k occurring is P(h k ), Assume that the transmission power of the status data packet is P 2 , and the channel fading coefficient is h within a certain coherence time k , then the signal-to-noise ratio of the control node receiving a status data packet is where d is the distance between the machine node and the control node, α is the path fading factor, and χ 0 is the shadow fading, and σ 2 is the noise power. Considering that the length of the data packet sent by the machine node is short, for this, the short data packet communication theory is adopted to characterize the transmission performance of the data packet. When the signal-to-noise ratio at the control node side is γ k , the amount of information transmitted is D nats, and when the data packet length is n channel uses, the probability ε k that the control node fails to decode the data packet is

[0029]

[0030] where the expression of the Q function in the above formula is x is the lower bound of the integral, and t is the integral variable

[0031] In the present invention, we use the age of information to characterize the freshness of the status information at the control node, which is defined as the difference between the current time slot and the generation time slot of the latest status data packet received by the control node. Define U t as the generation time slot of the latest status data packet received by the control node before time slot t. The AoI of the control node at time slot t can be expressed as

[0032] Δ t = t - U t

[0033] Only when the machine node sends a status data packet and it is successfully received by the control node, the age of information at the control node drops to 1. Otherwise, the age of information is incremented by 1, that is

[0034]

[0035] where Δ t+1 is the age of information of the control node before the start of the (t + 1)-th time slot

[0036] The present invention uses the average age of information to characterize the state update performance of the system, which is defined as

[0037]

[0038] where Υ is an integer approaching infinity

[0039] Step 4: The control node feeds back information

[0040] In the first time slot of a coherence time, the machine node sends a pilot, and the control node feeds back the obtained channel state information and the current age of information to the machine node. In the subsequent time slots of the same coherence time, as Figure 1 shown, since the channel fading coefficient remains unchanged within the same coherence time, the status data packet can be directly decoded by the control node without channel estimation. At the same time, the control node only feeds back the current age of information without feeding back the channel state information;

[0041] Step 5: State update strategy design

[0042] Before the start of the next time slot, the machine node determines the transmission strategy for the next time slot based on information such as the current channel state information, the current age of information of the control node, and the position of the next time slot within the coherence time. The present invention models the design problem of the state update strategy as a constrained Markov decision problem and obtains the optimal state update strategy through the method of associated value recursion. This Markov decision problem consists of four elements: state space, action space, state transition probability, and cost function, which are respectively represented as

[0043] State space S: The state before the start of time slot t can include: the AoI Δ at the control node t , the channel fading h t , and the position r of the time slot within a coherence time t . The state before the start of time slot t is represented as s t =(Δ t , h t , r t ).

[0044] Action space In each time slot, the machine node can take three possible actions: a t ∈{1, 2, 3}, where a t =1 means the machine node sends a pilot, a t =2 means the machine node sends a status data packet, and a t =3 means the machine node remains silent.

[0045] State transition probability P(s t+1 |s t , a t ) represents the probability that the state of time slot t + 1 is s t when the state of time slot t is s t and the action a t+1 is taken. The present invention first assumes that the machine node is in the first time slot, that is, r tWhen r = 1, in addition to transmitting pilots, it is also possible to choose to transmit status data packets or remain silent. By designing the state transition probability, the first time slot remains silent and finally the transmission of pilots is added to ensure that the obtained state update strategy meets the needs of state updates in non-ideal scenarios.

[0046] When r t = 1,

[0047] P(Δ t + 1, h t , r t + 1|Δ t , h t , r t , a t ) = 1, a t ∈ {1, 2, 3}.

[0048] When 1 < r t < t c ),

[0049] P(Δ t + 1, h t , r t + 1|Δ t , h t , r t , a t = 2) = ε t ,

[0050] P(1, h t , r t + 1|Δ t , h t , r t , a t = 2) = 1 - ε t ,

[0051] P(Δ t + 1, h t , r t + 1|Δ t , h t , r t , a t ∈ {1, 3}) = 1.

[0052] When r t = t c ),

[0053] P(Δ t + 1, h′, 1|Δ t , h t , r t , a t = 2) = ε t P(h′),

[0054] P(1, h′, 1|Δ t , h t , r t , a t = 2) = (1 - ε t )P(h′),

[0055] P(Δ t + 1, h′, 1|Δ t , h t , r t , a t = 3) = P(h′).

[0056] The probabilities of other state transitions are all 0. It should be noted that when r t = 1, the present invention is configured such that regardless of whether the machine node sends a status packet, the age of information increases by 1. In this case, the machine node sending a status data packet or a pilot in the first time slot will incur additional energy consumption but will not improve the status update performance. Therefore, the machine node will remain silent in the first time slot. Since the machine node must send a pilot in the first time slot to assist the control node in completing channel estimation, the present invention finally replaces the action in the first time slot with sending a pilot. In addition, since the age of information increases by 1 regardless of whether the machine node sends a pilot or remains silent in the first time slot, that is, adding a pilot will not affect the age of information, the solution obtained by this modeling method is consistent with the optimal solution.

[0057] Cost function : Considering the energy-constrained scenario, the present invention effectively improves the performance of status updates while satisfying the power consumption constraint. The cost function c((Δ, h, r), a) of the present invention is defined as the power consumption of the machine node and is defined as

[0058]

[0059] Given the initial state s 0 of the system and the status update strategy π, the expressions for the average age of information and the average power consumption are respectively

[0060]

[0061]

[0062] The optimization problem can be modeled as

[0063]

[0064]

[0065] where C maxLet \(P_0\) be the constraint value of the average power consumption. The strategy designed in the present invention refers to the actions that the machine nodes should take when the system is in a certain state.

[0066] To solve this constrained Markov decision problem, the present invention applies the Lagrange method to relax the original problem into an unconstrained Markov decision problem, defined as

[0067]

[0068] where \(\lambda>0\) is the Lagrange factor. For \(1_{\{\cdot\}}\), if the event in the brackets holds, this formula equals 1; otherwise, it equals 0.

[0069] Increasing \(\lambda\) can increase the weight of power consumption, thereby reducing the final power consumption of the obtained method. For the unconstrained Markov decision problem with a given \(\lambda\), there exists a deterministic method such that the following expression holds

[0070]

[0071] where the deterministic method means that for each state, the actions taken by the machine nodes are uniquely determined. is the minimum value of the Lagrangian function when the Lagrange multiplier is \(\lambda\), and \(f\) λ (s) is a bounded function, and \(s'\) is the next state of the system after taking action \(a\).

[0072] Introduce the "state-action" function \(V\), which is defined as

[0073]

[0074] According to the literature [L.I. Sennott, "Constrained average cost markov decision chains," Probab. Eng. Inf. Sci., vol. 7, no. 1, pp. 69–83, 1993.], the strategy that minimizes the Lagrangian function is

[0075]

[0076] To meet the power consumption constraint of the system, the present invention adopts an iterative method to find the optimal \(\lambda\). The specific steps are as follows: 1. Initialize a relatively large \(\lambda\) so that the obtained state update strategy satisfies the power consumption constraint, and simulate the power consumption when the Lagrange factor is \(\lambda\) 2. The update method of \(\lambda\) in the \(m\) -th iteration is where \(\upsilon\) m-1 is a sequence that gradually decreases to ensure the convergence of the algorithm, and simulate to obtain the power consumption 3. Repeat the above steps until or where τ is the convergence threshold of the algorithm. If it indicates that the current policy cannot meet the power consumption constraint for state update, and the optimal state update method is The specific solution implementation can be expressed as

[0077]

[0078] The system repeats steps 4 and 5. The control node feeds back the age of information and the channel state information to the machine node, and the machine node sends a state update data packet to the control node to reduce the age of information at the control node until the state update task ends.

[0079] Example:

[0080] Simulation parameter settings: The value of the channel fading coefficient is in the range of [0.64, 1], and its value interval is 0.03. The probability of each channel fading occurrence is the same. The transmission power of the pilot is P 1 = 0.2W, the transmission power of the state data packet is P 2 = 0.05W, and the average power consumption constraint is C max = 0.015W. The coherence time length is 40 time slots, the time slot length is n = 150 channel uses, the data packet information volume is D = 100 nats, the distance between the machine node and the control node is d = 200m, the path fading factor is α = 3.8, and the shadow fading is χ 0 = -50dB.

[0081] Figure 3 and Figure 4 show the structure of the optimal state update strategy obtained by the proposed method. As Figure 2 can be seen, when the number of time slots is small, that is, when the time slot is at the front position of a period of coherence time, the structure of the optimal method shows a threshold characteristic. For a certain channel fading, when the age of information exceeds a certain threshold, the machine node changes from silent to sending a state data packet. In addition, as the channel quality gets better, this threshold gradually decreases, because in the case of better channel quality, the control node can decode the state packet with a higher probability, thus reducing the energy waste caused by continuing to send data packets after decoding failure, so that the machine node can effectively reduce the age of information at the control node with limited energy. Figure 4 Fix the age of information at the current control node to 20 time slots, which means that the information freshness is very low. In this case, the machine node should send a state data packet to reduce the age of information as soon as possible. However, according to Figure 4It can be seen that when the channel fading is severe, at the end of a coherence time, the machine node will choose to remain silent. This is because when the channel quality is poor, the probability of decoding failure is relatively high. In this case, the machine node consumes energy to send status data packets, but may not be able to update the status information. Therefore, the machine node remains silent to save energy and then chooses to send status data packets when the channel quality is better at the next coherence time.

[0082] Figure 5 The average age of information of several different methods is shown. Among them, the random update method means that on the premise that the machine node sends a pilot in each coherence time and satisfies the power consumption constraint, the machine node randomly sends data packets or remains silent in each coherence time; the ideal channel state information method assumes that the control node knows the channel state in advance and there is no need for the machine node to send a pilot to it. At this time, the machine node can use the energy originally used to send the pilot to send status data packets, which can be regarded as an upper bound on the performance of the actual state update scenario. From Figure 5 it can be seen that the proposed method is superior to the random update scheme and greatly reduces the average age of information at the control node, which shows the effectiveness of the proposed method.

[0083] A pilot-assisted state update method under the condition of short data packet communication, which uses information such as the age of information, channel state information, and the position of the next time slot within the coherence time at the control node to adjust the state update behavior of the machine node in real time, and improves the freshness of the state information obtained at the control node. The machine node sends a pilot in the first time slot of the coherence time to help the system obtain the channel state information, and chooses to send status data packets or remain silent in the remaining time slots of a coherence time. The state update strategy design problem is modeled as a constrained Markov decision problem, and the method of associative value recursion is used to obtain the optimal state update strategy.

[0084] The description of the above embodiments is relatively specific and detailed, but only expresses 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 researchers and engineers 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 pilot-assisted status update method under short data packet communication conditions, characterized in that, it includes a machine node and a control node, each node is equipped with a single antenna; the machine node sends short data packets carrying status information to the control node, and the machine node periodically sends pilot signals to the control node to assist the control node in obtaining the current channel state information; including the following steps: Step 1: Determine the coherence time: The machine node determines the coherence time of the system according to its relative moving speed with the control node and the radio signal wavelength parameter; Step 2: Channel estimation: The machine node calculates the number of time slots passed within a coherence time, so as to send a pilot to the control node at the first time slot of each coherence time, and the control node estimates the channel state information of the current coherence time according to the pilot; Step 3: Perform status update: After the machine node sends the pilot, within the same coherence time, it sends a status data packet to the control node or remains silent. If the machine node sends a status data packet and it is successfully received by the control node, the information of the control node is updated and the age of information is reduced to 1, otherwise the age of information increases by 1; Step 4: Control node feedback information: At the first time slot of a coherence time, the machine node sends a pilot, and the control node feeds back the obtained channel state information and the current age of information to the machine node. In the subsequent time slots of the same coherence time, since the channel fading remains unchanged within the same coherence time, at this time the control node only feeds back the current age of information without feeding back the channel state information; Step 5: Status update strategy design: Before the start of the next time slot, the machine node determines the sending strategy for the next time slot according to the current channel state information, the current age of information of the control node, and the position information of the next time slot within the coherence time:

1. Send a pilot, 2. Send a status data packet, 3. Remain silent; The system repeats steps 4 and 5 until the status update task ends.

2. The status update method according to claim 1, characterized in that, in order to characterize the performance of the status update system and reflect the improvement of the system status update performance, the adopted metric is the average age of information of the control node.

3. The status update method according to claim 1, characterized in that, the data packets sent by the machine node are short, and the machine node can send multiple data packets within a coherence time.

4. The status update method according to claim 1, characterized in that, at the first time slot of each coherence time, the machine node sends a pilot to the control node to help the control node complete channel estimation.

5. The status update method according to claim 1, characterized in that, Before the start of the next time slot, the machine node determines the transmission strategy for the next time slot based on the current age of information of the control node, the current channel state information, and the position of the next time slot within a coherence time. The state space of the system is Before the start of time slot t, the state s of the system t is s t =(Δ t , h t , r t ), where Δ t is the current age of information of the control node, h t is the current channel fading coefficient, and r t represents that time slot t is the r t -th time slot within a coherence time.

6. The status update method according to claim 5, characterized in that, Δ t The maximum value that can be taken is Δ max , Δ t The range of values of Δ t ∈ [1, Δ max and wherein, is the set of positive integers, and the value of Δ max is large enough so that during the actual state update process, Δ t cannot exceed Δ max ; the set of channel fading coefficients is The values of the channel fading coefficients are discrete. When the number of sampling values of the channel fading is large enough, this discrete channel fading model can be approximated as the actual continuous channel fading model; r t The range of values of r t ∈ [1, t c and wherein, t c is the total number of time slots within a coherence time; by setting a maximum value for the age of information and discretizing the channel fading, the state space of the system is made finite.

7. The status update method according to claim 1, characterized in that, Before the start of each time slot, the machine node has three alternative strategies, and the action space of the system is The strategy adopted for the t-th time slot is denoted as a t ∈ {1, 2, 3}, where a t = 1 means that the machine node sends a pilot in the t-th time slot, a t = 2 means that the machine node sends a status data packet in the t-th time slot, a t = 3 means that the machine node remains silent.

8. The status update method according to claim 7, characterized in that, The status information at the control node is updated only when the machine node sends a status data packet and it is successfully received by the control node, and the age of information is reduced to 1; otherwise, the age of information is incremented by 1, that is where Δ t+1 is the age of information at the control node before the start of the (t + 1)-th time slot.

9. The status update method according to claim 1, characterized in that, the optimization problem of the status update strategy is modeled as a constrained Markov decision problem, and the optimal status update strategy is obtained by using the method of associated value recursion.

Images