Hybrid time slot allocation method and communication method for power line carrier communication

Abstract

The invention discloses a mixed time slot allocation method and a communication method for power line carrier communication. The mixed time slot allocation method comprises the following steps: dividing each phase of data frame trained in turn into a fixed time slot, a pre-allocation stage and a reserved dynamic time slot; wherein the fixed time slot is determined by the network node number and the node type; based on the previous business volume of each member node, Kalman filtering is adopted for prediction to obtain the task volume of the corresponding member node at the current moment; converting the predicted task load at the current moment into the total time slot demand number of each member node; and obtaining the dynamic reservation time slot number of each member node according to the total time slot demand number of each node and the fixed time slot number of each member node determined by the fixed time slots in the allocation time slots of the pre-allocation stage. The remaining time slots are used for dynamic reservation allocation according to the prediction result, and remaining time slot resources are flexibly allocated on the premise that basic time slot requirements are guaranteed.

Description

Technical Field

[0001] The invention relates to the technical field of electrode carrier communication, and in particular to a power line carrier phase switching method. Background Art

[0002] In power line carrier communication, there are three common time slot allocation methods, namely fixed time slot allocation, contention time slot allocation and dynamic time slot allocation.

[0003] Fixed time slot allocation: In fixed time slot allocation, each time slot is fixedly occupied by a participating member. When this time slot arrives, only this member has the right to send data, and other members can only receive. In this mode, all time slot arrangements have been determined before transmission. Each node can transmit data when its own time slot arrives. There is no need to judge the service. The mechanism is relatively simple, but the transmission in this mode is too fixed and the flexibility is very poor, which may lead to waste of time slots and failure to transmit high-priority services in a timely manner.

[0004] Contention time slot allocation: The present invention relates to the field of electrode carrier communication technology, and specifically to a power line carrier phase switching method.

[0005] In the competitive time slot access mode, each network member can occupy the time slot resources in the time slot pool by competing according to its own demand for time slots. It is generally used in big data services where the dedicated access mode cannot meet the transmission continuity requirements. Its advantage is that high-priority services can quickly occupy transmission channel resources, ensuring the delay requirements; its disadvantage is that collisions caused by competition may cause the success rate of information transmission to decrease, especially when there are many nodes competing for time slots.

[0006] Dynamic timeslot allocation:

[0007] In the MAC protocol, if the allocation method of communication resources changes with the user's information transmission needs, then this allocation method is called a dynamic allocation method. The dynamic allocation method has higher resource utilization efficiency because it can show adaptability to signal transmission needs. The design of the power line carrier three-phase carrier communication system allows each access node to request the CAC node to dynamically allocate corresponding time slot resources for it according to its own needs. Each node will be unified according to its own needs and the needs of other nodes for unified planning. If the total time slot demand exceeds the total time slot resources, the respective time slot demand will be reduced in a certain proportion. This access mode has the best flexibility, but it is necessary to collect statistics and predictions on the business conditions of each node in the three-phase carrier so as to reallocate resources before the next cycle. The advantage is that the time slot resources can be fully utilized, the upload delay of high-priority services can be reduced, and the flexibility is high and the real-time performance is good; the disadvantage is that nodes with low service priority may not be allocated time slots for transmission, and fairness is difficult to guarantee. Summary of the invention

[0008] The technical problem to be solved by the present invention is to provide a hybrid time slot allocation method based on network scale and service prediction, so as to realize flexible allocation of remaining time slot resources under the premise that basic time slot requirements are guaranteed.

[0009] In order to solve the above technical problems, the technical solution adopted by the present invention is: A hybrid time slot allocation method for power line carrier based on network scale and service prediction, comprising: The data frame of each phase of round-robin training is divided into fixed time slots, pre-allocation phase and reserved dynamic time slots; the fixed time slots are determined by the number of network nodes and node types; the pre-allocation phase is divided into three parts: request time slots, allocation time slots and broadcast time slots; the reserved dynamic time slots access each node for data transmission according to the allocation results of the broadcast time slot broadcast; Based on the previous business volume of each node, Kalman filtering is used to predict the task volume of the corresponding node at the current moment; the predicted task volume at the current moment is converted into the total number of time slot requirements of each node; In the allocated time slots of the pre-allocation phase, the number of dynamically reserved time slots of each node is obtained according to the total number of time slot requirements of each node and the number of fixed time slots of each node determined by the fixed time slots.

[0010] Compared with the prior art, the present invention has the following beneficial effects: The present invention proposes a method combining fixed allocation and dynamic allocation of time slot resources based on business prediction. First, a part of the time slots are fixedly allocated to the nodes of each phase, and the remaining time slots are used for dynamic reservation allocation according to the prediction results. The remaining time slot resources are flexibly allocated under the premise that the basic time slot demand is guaranteed.

[0011] Because there are both stable periodic services and sudden emergency services in the power line carrier data link, for sudden services that require dynamic time slot allocation, the present invention adopts the Kalman method to predict sudden services, and makes real-time predictions on the service volume of the next cycle by observing the service volume and service status of the node at the current and past moments. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a comparison chart of the actual data collected and the predicted data by the Mann filter; Figure 2 It is a frame structure design diagram; Figure 3 It is a schematic diagram of dynamic priority scheduling; Figure 4 Average service latency graph for different network nodes; Figure 5 Average delay of emergency services at different network nodes. DETAILED DESCRIPTION

[0013] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0014] The invention proposes a hybrid time slot allocation method for power line carrier based on network scale and service prediction. The algorithm combines node fixed time slot allocation with dynamic time slot reservation mechanism, uses Kalman method to predict the service volume and service priority distribution of each node on three phases in the next cycle, CAC node collects time slot request information of routing control nodes on each phase and performs dynamic time slot reservation allocation in combination with its service priority, then designs carrier switching algorithm according to the scale of three-phase access network, time slot allocation of each phase and channel quality, finally compares the delay and throughput of the original scheme with the proposed scheme and performs simulation verification.

[0015] The communication requirements in power carrier communication include the communication requirements for marketing business and the communication requirements for distribution network operation business, see Table 1 and Table 2.

[0016]

[0017] The hybrid time slot allocation method based on network scale and service prediction provided by the present invention first predicts the communication demand through the Kalman adaptive filtering algorithm. The Kalman adaptive filtering algorithm is as follows: Initial conditions:

[0018]

[0019] in, For the current moment The sampling volume; is the sampling business volume at the current time 1; is the mean of the sampled business volume before the current time 1; represents transpose; It means to find the mean value; The observation vector sequence is , the known parameters are: state transfer matrix , observation matrix , process noise correlation matrix ,Observation noise correlation matrix . The calculation is as follows: Predict the next state value based on the previous state of the system:

[0020] Error (difference between observed and predicted values):

[0021] Gains:

[0022] Based on the predicted and measured values, the optimal estimate of the current state n can be obtained:

[0023] The forecast covariance at time n is:

[0024] State covariance at time n:

[0025] Represents the state covariance matrix at time n.

[0026] The Kalman filter is a linear discrete-time finite-dimensional system. The estimation performance of the Kalman filter is: it makes the correlation matrix of the state estimation error after filtering The trace of is minimal. This means that the Kalman filter is the state vector The linear minimum variance estimate of .

[0027] For example, the nodes on the three phases of the power line carrier generate burst traffic using the Weibull distribution Simulation. Assuming that each node has 100 types of services, we use the traffic data of the node in the first 20 cycles as known data, use the Kalman filter to predict the service value of the subsequent cycle, and compare the actual data collected with the predicted data. The prediction effect is as follows: Figure 1 shown.

[0028] Depend on Figure 1 It can be seen that due to the instability of the business, there is a certain deviation between the prediction result and the real data, but it is basically close to the real data, and the lag between the predicted value and the real value is small and can be ignored. Therefore, Kalman filtering can well realize business prediction.

[0029] Frame structure design like Figure 2 As shown, the frame structure is designed to be composed of three parts: fixed time slots, allocation phase and reserved dynamic time slots.

[0030] In the fixed time slot stage, the corresponding algorithm completes the allocation of fixed time slots. , T is the given parameter value, and N is the number of network nodes. The pre-allocation phase is divided into three parts: requesting time slots, allocating time slots, and broadcasting time slots. This process can be completed quickly by CAC nodes executing related algorithms. Therefore, the time slot demand is very small, and only a small number of time slots need to be allocated. The specific implementation of the time slot demand calculation and dynamic reservation algorithm in this phase refers to the dynamic time slot allocation algorithm above. After the pre-allocation of time slots and the broadcasting of allocation results are completed, the reservation dynamic time slot execution phase is entered. Each phase node can send services when the allocated time slot arrives according to the marked time slot status.

[0031] Algorithm Flow Fixed time slot allocation phase At this stage, some time slots in a time frame cycle are used to be fixedly allocated to each node in the three phases. Fixed time slots are used to provide network access, clock synchronization, and service data that have not been transmitted in the previous time frame cycle for each node. The number of fixed time slots will increase or decrease in each frame cycle according to the number of nodes accessing the network in the current cycle, that is, the allocation of fixed time slots in each frame cycle is adjustable. The fixed allocation method of time slots adopts uniform allocation. Uniform allocation is to extract time slots at equal intervals in a network cycle and allocate the extracted time slots to each node. The extracted time slots are called slot blocks. This can achieve a reasonable allocation of the number of time slots and a uniform planning of the time slot positions, and the allocation is performed using the traditional binary tree algorithm.

[0032] Dynamic timeslot allocation In the request phase, the service reservation time slot requirements of each node on the three phases A / B / C are calculated by the Kalman prediction results of the number of data packets in each phase in the past and current time and the emergency situation. The prediction method is the same as the Kalman filter prediction mentioned above. If the fixed time slots that the node has obtained cannot meet its time slot requirements, the excess time slots need to be met through reservation. Therefore, in the time slot request phase, each node in the three phases converts the service prediction results into the required number of time slots and sends a time slot request packet to the CAC node. After the time slot request phase ends, the CAC node will obtain the time slot requirement table of all nodes on the network.

[0033] In the pre-allocation phase, the head-end node, as the network controller, is responsible for receiving time slot requests, completing dynamic time slot allocation, and broadcasting the allocation results. Assume that the total number of time slots required for various priority services on the nodes on the A / B / C three-phase is , Indicates Root phase line The total number of time slots required by members (arranged in descending order by node number) in one frame. In the fixed time slot allocation stage, the total number of time slots allocated to the nodes on each phase is , then the number of reservation time slots required is . After CAC collects the time slot demands on each phase, it allocates them from high to low according to the predicted service priority of each node, and takes into account the characteristics of power line carrier transmission (in the same time slot, the CAC node can only communicate with one phase node among the three phases; it cannot communicate on one phase for a long time), and designs a reasonable allocation of time slots corresponding to each phase node. If the number of demands of each phase node exceeds the total number of reserved time slots, time slot resources are first allocated to the nodes corresponding to the high-priority services, and the remaining resources are allocated proportionally to low-priority users. The allocation of dynamic reserved time slots is only valid within the current time frame period. After the fixed time slot of the next time frame period ends, the CAC node needs to reallocate the number of reserved time slots based on the new round of time slot demand information.

[0034] The specific algorithm for time slot pre-allocation is as follows: Respectively represent Root phase line The total number of time slots required by members (arranged from large to small by node number) in one frame, the number of time slots already obtained in the fixed allocation phase, and the number of time slots that need to be dynamically reserved. Respectively represent the total number of time slots in a time frame period, the number of time slots occupied by fixed allocation, and the number of time slots remaining for dynamic reservation. After the CAC node obtains the time slot requirement table required by each phase node, it can start allocating time slots, which can be divided into the following two cases: (1) If After the total number of time slots required for each phase line is allocated according to the predicted service priority order, the head-end node allocates the total time slots to all members in proportion. Since the time slot requirements of each node are small in this case, the time slots are not completely used up. In order to prevent the system from wasting bandwidth and causing the subsequent service delay to be too large, the current time frame is ended and the fixed time slot phase of the next time frame cycle begins. The CAC node is responsible for the broadcast task of the entire network.

[0035] (2) If . This means that the remaining time slots are not enough for all members to be allocated all the required time slots. If a completely fair method is used, the QoS of the highest priority service cannot be guaranteed; if the time slots are allocated in order of priority, low-priority services will be unable to send data for a long time. Therefore, a reasonable allocation method is needed to ensure that the delay of high-priority services is not too large while taking fairness into account. First, set the corresponding priority weights for the two services (marketing and distribution network) of each phase node. , K and M represent the priority types of two types of services. is the upper limit of the proportion of the highest priority business. The lower limit of the time slot ratio occupied by the service with the lowest priority. The two services use different frequency bands for data interaction and therefore do not affect each other, but the weight of the former is higher. Therefore, we need to reasonably coordinate the time slot allocation corresponding to the priority weights of the two services at each node. The constraints for dynamic reservation of time slots are:

[0036] The constraints of dynamic reservation time slots indicate that the service quality of high-priority services must be guaranteed. If the fixed time slots allocated to high-priority services are sufficient, dynamic time slots can be obtained by other priority services. A smaller value can be set so that it does not compete with high-priority services for too many transmission time slot resources and cannot guarantee QoS requirements. However, this fair approach will still make the time slots insufficient, so the excess services will be scheduled to the fixed time slots or reserved time slots of the next frame period. The CAC node reasonably designs the switching strategy based on the number of time slots allocated to the nodes on each phase. After the reserved time slots are allocated in proportion, the status of each time slot needs to be marked. Each member clarifies the time slots they occupy based on the marking situation, and then completes the designated sending task when the dynamic reserved time slot arrives.

[0037] Switching strategy The first step is to access a fixed time slot for each phase in rotation (fixed time slot allocation can ensure the basic service transmission of each phase). The rotation access time is adjusted in real time according to the network scale (large, medium, small) to ensure the basic time slot requirements of the three phases.

[0038] In the second step, after the prediction algorithm dynamically allocates the reservation time slot, the access time is dynamically adjusted based on the allocation result and the fixed training time slot. Taking into account the switching delay and the characteristics of power line carrier transmission (data cannot be transmitted on one phase for a long time), the switching strategy should achieve a balance between the two.

[0039] Step 3: Reasonably select three-phase access according to the quality of the three-phase channels in each cycle. For example: In this cycle, the channel quality of phase A is relatively good, and as much data interaction as possible should be allocated to phase A on the basis of ensuring the correct transmission rate. The nodes on phase B / C can send signals to the nodes on phase A through radiation or wireless to report data (the number of hops increases, and the delay becomes relatively large), or detect the channel quality of other phases after a period of time, and then change the access time.

[0040] Taking the above into consideration, the overall system latency can be reduced while ensuring the latency of various priority services.

[0041] Performance Analysis

[0042] The reserved time slots will be dynamically adjusted according to the change of the time frame period. Therefore, the present invention adopts the method of allocating the power line carrier three-phase nodes according to the priority service situation. The time slots are continuously allocated according to the priority level, and the time slot positions are also continuous. Figure 3 As shown, each colored time slot segment represents the time slot occupancy of a node until the time slot allocation of low priority services is completed, and the fixed time slot of the next time frame cycle begins. The specific analysis is as follows: The service demand time slots of the three-phase routing aggregation node collected by the CAC node are allocated to the respective priority queues in the three phases according to the service priority. Assuming that the total number of nodes in the three phases is N, the services generated by users follow the Poisson distribution, and the rates of the two services from high to low priority are , K and M represent the priority types of two types of services. is the upper limit of the proportion of the highest priority business. is the lower limit of the proportion of time slots occupied by the lowest priority service. The first service has a higher priority than the second service. The node with a high service priority gets the opportunity to send first, with the smallest delay. The process of multiple time slots being occupied and served continuously can be regarded as a simple service process in queuing theory. It is assumed here that the rate at which the messages sent in each time slot are processed by the system is the same, that is, the data processing capacity of the CAC nodes in each time slot is unchanged. Through the above analysis, we can get the delay of the nodes on each phase, for example: the priority on phase A is Business delay only includes the following four parts: (1) The processing time required for the higher priority business being served (2) Waiting time of phase A caused by receiving service on B / C (3) Processing time required for this queue (4) Switching delay If the remaining dynamic time slots of the higher priority service in phase A are , the priority on phase A is The total business volume (i.e., number of messages) of all members is , the receiving service time of each message is (other processing time is not included), Indicates the delay required for switching once, then the priority of phase A is Average queue waiting time of the node for:

[0043] Among them, J and G represent The number of times that phases B and C access the CAC node for data transmission before the priority service is served; The probability that the system is busy; The number of remaining dynamic time slots for the A-phase higher priority services The average value of The time required for phase B to access the CAC node for the jth data transmission; The time required for phase C to access the CAC node for the g-th data transmission.

[0044] Similarly, the service priorities of phases B and C are: The average queue waiting time is Since the system predicts the services of each phase node in the next time frame period in the dynamic reservation allocation phase, it is assumed that all time slots in the dynamic execution phase have data to send, so the probability of the system being busy is . Priority To simplify the analysis, it is stipulated here that the service arrival follows the Poisson distribution, so the priority can be obtained from Little's theorem as The average number of messages in each phase is:

[0045] The processing time for each message in the system is the same, both ,therefore

[0046] If the current power line carrier transmission is processing a higher priority message, is the service time after the CAC node switches phase A once. From the above, we can see that the service time of the message follows a fixed-length distribution. Therefore, the remaining processing time of the higher priority has a stable distribution, and the mean is:

[0047] In summary, it can be deduced that the priority on phase A is The average queue waiting time of the node is:

[0048] Therefore, the priority of phase A is The total service delay is the sum of the queuing delay and the processing delay:

[0049] Similarly, the time delay on phases B and C can be obtained as:

[0050] Simulation Verification The frame structure design, service conditions, and service size settings of this article are as follows: Slot length / ms: 20 The length of the time slots contained in the time frame: 32 Bandwidth: 125kbps Burst traffic ON distribution: Weibull (5,3) Distribution of sudden business OFF: Poisson (5) Packet generation interval: 0.01s Packet length (small, medium, large) / Byte: 32, 64, 128 Number of bits transmitted per time slot / Byte: 960 The time frame length is 640ms, so the number of frames transmitted per second is 1 / 0.640. The system throughput is the product of the number of time slots per frame, the number of bits transmitted per time slot, and the number of frames per second. The theoretical saturation throughput under the configuration in this article is: 32×120×8 bit ×1 / 0.640 =48000 bps. The delay size is directly related to the business volume setting and the priority level. In actual communication, the throughput will decrease due to the transmission of non-data packets such as reservation requests and the overhead of the packet header size.

[0051] The simulation diagrams of the overall average delay and the average delay of emergency events of the three algorithms of fixed allocation method, competitive access method, and fixed + dynamic allocation are shown in the figure. Figure 4 , Figure 5 As shown: from Figure 4 , Figure 5It can be seen that as the network scale increases, the average latency of the service increases. Among them, the average latency of the fixed time slot allocation method under large-scale network nodes is greatly improved, and the performance is relatively poor. This is because when there are many network nodes, fixed time slot allocation will waste more time slots and cause the service to not be uploaded in time. The performance of competitive time slot allocation is relatively good under small-scale nodes, but with the increase of network nodes, the probability of service transmission competition collision is greatly increased, which reduces the transmission success rate, and thus requires more time slots to retransmit data. The fixed + dynamic allocation time slot method can adapt well to the elastic changes in network scale. Although the latency will also increase with the increase in the number of network nodes, it is relatively small.

Claims

1. A hybrid time slot allocation method for power line carrier based on network scale and service prediction, characterized in that: include: Divide the data frame of each phase polling into fixed time slots, pre-allocated phases and reserved dynamic time slots; The fixed time slot is determined by the number of network nodes and the node type; the pre-allocation phase is divided into three parts: request time slot, allocation time slot and broadcast time slot; the reserved dynamic time slot is connected to each member node for data transmission according to the allocation result of the broadcast time slot broadcast; Based on the previous business volume of each member node, the Kalman filter is used to predict the task volume of the corresponding member node at the current moment; the predicted task volume at the current moment is converted into the total number of time slot requirements of each member node; In the allocated time slots of the pre-allocation phase, the number of dynamically reserved time slots for each member node is obtained according to the total number of time slot requirements of each node and the fixed number of time slots of each member node determined by the fixed time slots.

2. A hybrid time slot allocation method for power line carrier based on network scale and service prediction according to claim 1, characterized in that: Based on the previous traffic volume of each member node, the Kalman filter is used to predict the task volume of the corresponding member node at the current moment; in the step of converting the predicted task volume at the current moment into the total number of time slot requirements of each member node, the method of using the Kalman filter to predict the traffic volume of each member node to obtain the task volume of the corresponding node at the next moment is: Predict the next state value based on the previous state of the system: in, Predicting the value of the traffic volume at time n based on the value of the sampled traffic volume at time n-1; is the state transfer matrix of the Kalman filter; To predict the value of the traffic volume at time n-1 based on the value of the adopted traffic volume before time n-1; Compute the difference between the observed and predicted values: in, is the difference between the observed value and the predicted value; For The observation vector at time instant; is the observation matrix of the Kalman filter; Calculate the gain: in, To calculate the gain; is the predicted covariance from time n-1 to time n; is the observation matrix of the Kalman filter at time n; Observation noise correlation matrix; Based on the predicted and measured values, the optimal estimate of n at the current moment is obtained: in, For the current moment Forecasted business volume; The forecast covariance at time n is: in, is the predicted covariance transferred from time n-1 to time n-1; is the process noise correlation matrix; State covariance at time n: in, Represents the state covariance matrix at time n.

3. A hybrid time slot allocation method for power line carrier based on network scale and service prediction according to claim 1, characterized in that: The Kalman filter is used to predict the traffic volume of each member node to obtain the task volume of the corresponding node at the next moment. In the step of converting the predicted task volume into the total number of time slot requirements of each member node, the predicted task volume is converted into the total number of time slot requirements of each member node: in, is the total number of time slots required for each member node; The transmission rate of each member node is a given parameter.

4. A hybrid time slot allocation method for power line carrier based on network scale and service prediction according to claim 1, characterized in that: The number of time slots for fixed time slot allocation is: in, The number of time slots allocated to each member node in the fixed time slot phase of a frame; T is a given parameter value, and N is the number of network member nodes.

5. A hybrid time slot allocation method for power line carrier based on network scale and service prediction according to claim 4, characterized in that: In the allocation time slot of the pre-allocation phase, the number of dynamic reserved time slots of each node is obtained according to the total number of time slots required by each member node and the number of fixed time slots of each member node determined by the fixed time slots. The number of dynamic reserved time slots of each node is obtained as follows: in, The number of time slots that each node member needs to dynamically reserve in a frame.

6. A power line carrier communication method, characterized in that: include: Obtaining the number of allocated time slots for each node based on the hybrid time slot allocation method of the power line carrier based on network scale and service prediction according to any one of claims 1-5; According to the number of allocated time slots, data is transmitted to each access node; According to the quality status of the three-phase channels in each cycle, the phase with the best quality among the three phases is selected for access and data transmission is completed.

7. The power line carrier communication method according to claim 6, characterized in that: After the CAC node obtains the time slot requirement table required by each phase node, it can start allocating time slots, which can be divided into the following two cases: (1) If , Indicates the number of remaining dynamically reservable time slots in a time frame period. Indicates Root phase line The number of time slots that each member needs to dynamically reserve in a frame; after allocating the total number of time slots required for each phase line according to the predicted service priority order, the head-end node allocates the total time slots to all members in proportion; the current time frame ends and the fixed time slot phase of the next time frame cycle begins; the CAC node is responsible for the broadcasting task of the entire network; (2) If First, set the corresponding priority weights for the two services of each phase node , k and m represent the priority types of two types of services, is the upper limit of the proportion of the highest priority business. The lower limit of the ratio of time slots occupied by the lowest priority service; Each phase is trained in a fixed time slot, and the training access time is adjusted in real time according to the network scale to ensure the basic time slot requirements of the three phases; After the prediction algorithm dynamically allocates the reservation time slot, the access time is dynamically adjusted based on the allocation result and the fixed round-robin time slot; According to the quality status of the three-phase channels in each cycle, one of the three phases is selected for access.

Images