Method and system for interference management in heterogeneous domain network based on interference-aware prediction and time slot allocation

Abstract

The application discloses a kind of based on interference perception prediction and time slot allocation's isomer domain network interference management method and system, belong to wireless communication technical field, method includes: coordinator in data transmission stage real-time observation current channel interference power, and in non-active period by channel hopping is carried out sparse sampling, update each channel's time slot interference weight and node propagation factor;Establish interference prediction model, predict the average interference intensity of each channel in future several frames;Inter-network game model is constructed, so as to select the communication channel that makes inter-network game model benefit function maximization;According to the prediction result of interference prediction model, local observation and time slot interference weight, obtain the interference power prediction value of each time slot in channel;Based on time slot interference power prediction value and node propagation factor, time slot is allocated for sensor node.The application is applicable to dynamic isomer domain network coexistence interference scene, and inter-network interference can be effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and in particular to a method and system for managing interference in heterogeneous area networks based on interference perception prediction and time slot allocation. Background Technology

[0002] Wireless Body Area Networks (WBANs) are low-power, lightweight communication systems designed to collect and monitor human physiological data (such as blood pressure, blood sugar, and electrocardiogram) using wireless sensor nodes. In standard deployments, WBANs typically employ a star topology, consisting of several wireless sensor nodes and a central coordinator: the sensor nodes collect physiological data and transmit it to the coordinator, which then forwards the received data to medical centers or remote servers, supporting applications such as health monitoring and disease early warning.

[0003] However, when multiple body area networks (WBANs) coexist in a confined space, the probability of co-frequency transmission increases significantly, easily leading to severe inter-network interference. This interference significantly reduces the signal-to-noise ratio (SINR), resulting in decreased system throughput, increased packet loss rate, and exacerbated energy consumption of sensor nodes. Especially in critical application scenarios such as healthcare, communication reliability is directly related to the validity and timeliness of monitoring data. How to achieve interference suppression and efficient resource scheduling in complex dynamic environments has become a core challenge facing current WBAN technology.

[0004] Existing technologies have failed to effectively balance the relationship between interference suppression, resource utilization efficiency, and energy consumption control, making it difficult to meet the differentiated transmission needs in dynamic heterogeneous area network coexistence scenarios. Therefore, there is an urgent need for a high-precision, low-overhead, and highly adaptable interference management method. Summary of the Invention

[0005] The purpose of this invention is to provide a heterogeneous area network interference management method and system based on interference perception prediction and time slot allocation, which aims to solve or improve at least one of the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A heterogeneous area network interference management method based on interference sensing prediction and time slot allocation includes:

[0008] S1. The coordinator observes the interference power of the current channel in real time during the data transmission phase, and performs sparse sampling through channel transitions during inactive periods to update the time slot interference weights and node propagation factors of each channel.

[0009] S2. Establish an interference prediction model to predict the average interference intensity of each channel in the next few frames.

[0010] S3. Construct an inter-network game model, treating each heterogeneous body area network as a game participant, evaluating the benefits of each channel through prediction results, conducting non-cooperative games among multiple coexisting body area networks, and selecting the communication channel that maximizes the benefit function of the inter-network game model.

[0011] S4. Based on the prediction results of the interference prediction model, local observations, and time slot interference weights, obtain the predicted interference power value for each time slot in the channel.

[0012] S5. Based on the predicted value of time slot interference power and the node propagation factor, calculate the sensor node's preference ranking for time slots and the time slot's preference ranking for sensor nodes, respectively, allocate time slots to sensor nodes, and perform local time slot rematching operation on the interfered node when interference is detected.

[0013] Further, step S1 specifically includes:

[0014] S1.1 The coordinator observes the current working channel in real time during the data transmission phase and collects the interference noise power of the current channel while receiving data from sensor nodes.

[0015] S1.2 The coordinator switches to other channels during time slots and sleep periods when no sensor nodes are transmitting data, collects channel energy, and obtains the interference noise power of the current channel;

[0016] S1.3 The coordinator records the interference noise of each channel, calculates the proportion of interference noise in each time slot within a preset sliding window, and performs exponential weighted recursive calculation on the interference noise proportion to obtain the time slot interference weight of each channel in different time slots.

[0017] S1.4 The coordinator counts the data packet transmission success rate of each sensor node within a preset sliding window, and at the same time collects and performs a sliding average processing on the received power of each node. The node propagation factor is obtained by dividing the received power after the node sliding average by the average received power of all sensor nodes in the network.

[0018] Furthermore, S2 specifically refers to:

[0019] S2.1 Based on the collected multi-channel interference data, calculate the average interference intensity of each channel in each frame, define an interference state vector including average interference intensity and interference power change rate, introduce three interference modes: stable interference, moderately variable interference, and burst interference, and construct a switching linear system.

[0020] S2.2. Create a corresponding state transition model and observation model for each interference mode. Fuse the state transition models and observation models of the three interference modes through the Markov transition matrix. Perform interactive mixing, mode filtering, likelihood and mode probability update, and fusion output operations in sequence to complete the global state estimation of interference for each channel.

[0021] S2.3 Based on global state estimation, predict the average interference intensity of each channel in the next few frames, and derive the signal-to-interference-plus-noise ratio and packet loss probability of each channel in the next few frames based on the average interference intensity.

[0022] Furthermore, S3 specifically refers to:

[0023] S3.1 Treat each heterogeneous wireless body area network as an independent game participant, construct an inter-network non-cooperative game model, define a utility function with channel throughput and transmission energy consumption as the core, calculate the utility function value of each channel based on the signal-to-interference-plus-noise ratio and packet loss probability obtained from interference prediction, and use it as the payoff value of each channel.

[0024] S3.2 Statistically analyze the actual throughput of historical communications. If the preset throughput threshold is not reached, use the Boltzmann selection strategy to generate the selection probability distribution of each channel and select the communication channel according to the probability. If the threshold is reached, continue to use the original communication channel.

[0025] S3.3 Real-time statistics of packet loss rate of each sensor node. When the number of interfered nodes that meet the preset packet loss rate threshold reaches the set lower limit, the re-game process is immediately triggered to re-execute the channel benefit assessment and non-cooperative game and select a new communication channel.

[0026] Furthermore, S4 specifically includes:

[0027] First, a game of channel switching is played. Then, the average interference intensity output by the interference prediction model is multiplied by the time slot interference weight of each time slot to obtain the predicted interference power value of each time slot of the channel.

[0028] Furthermore, S5 specifically includes:

[0029] S5.1 Calculate the required time slot quota for each sensor node based on the amount of data to be uploaded, the data transmission rate, and the fixed length of the time slot for each sensor node.

[0030] S5.2. Based on the remaining tolerable delay of sensor nodes, candidate time slots are screened, and the preference ranking of each sensor node for time slots is generated by layering according to the signal-to-interference-plus-noise ratio and combining the time proximity.

[0031] S5.3 Classify the nodes according to the temporal continuity characteristics of the time slots allocated to the sensor nodes, and generate a preference ranking of the sensor nodes for each time slot by combining the service priority of the fusion node and the dynamic utility value of the node propagation factor.

[0032] S5.4. Based on bidirectional preference sorting, the initial time slot allocation is completed. Nodes that do not meet the time slot quota initiate a matching request to the best time slot in their preference list. The time slot selects a better node according to its own preference list to complete the matching update. The rejected nodes initiate requests to the next preferred time slot in turn until all nodes meet the quota or there are no available time slots.

[0033] S5.5. Monitor the communication status of sensor nodes in real time and perform local time slot rematching operation on the interfered nodes.

[0034] Further, step S5.5 specifically includes:

[0035] The communication status of sensor nodes is monitored in real time. When interference is detected, the time slot matching relationship of the undisturbed nodes is frozen, and only the matching relationship of the interfered nodes is released and their bidirectional preference ranking is updated. The interfered nodes re-initiate matching requests based on the updated preference list. The time slots are only locally compared and matched between the currently occupying nodes and the newly requesting nodes, thus realizing local time slot rematching of the interfered nodes.

[0036] The present invention also provides a wireless body area network system that applies any of the above-mentioned interference sensing prediction and time slot allocation-based heterogeneous body area network interference management methods, comprising a plurality of heterogeneous wireless body area networks, each of which includes a coordinator and a plurality of sensor nodes.

[0037] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0038] 1. This invention introduces the IMM algorithm to construct an interference evolution SLDS system, which runs Kalman filters corresponding to three interference modes: steady-state, slowly varying, and bursty. Through real-time model probability updates, it quickly identifies and predicts interference change trends. Human dynamic movements (limb swinging, torso rotation, etc.) cause drastic fluctuations in WBAN channel path loss and shadow fading, posing challenges to channel selection and anti-interference. Single Kalman filter prediction or location-based interference prediction does not consider scenarios such as human turning and occlusion, making it difficult to adapt to human dynamic movements and resulting in insufficient prediction accuracy. Therefore, this invention combines multi-mode capture of social behavior and occlusion-induced abrupt interference to improve prediction accuracy.

[0039] 2. This invention constructs a channel-level interference evolution model, combines the coexistence history of nodes with interference statistical characteristics, identifies the persistence and scope of interference, and introduces a time slot-level local reallocation strategy. In the early stage of interference, time slot scheduling is adjusted only for affected nodes to avoid global channel switching. Channel-level reselection is only triggered when it is predicted that the interference will continue to worsen and local repair cannot meet the transmission requirements.

[0040] 3. This invention utilizes only energy observations, packet loss statistics, and local prediction results available from the coordinator side for channel selection and time slot allocation, avoiding reliance on external network private scheduling information and improving project deployability. Existing distributed anti-interference strategies rely on neighbor networks to obtain information about the channels used by neighbors in the body area network or the time slot activation status of internal sensor nodes. However, in practical systems, such information is difficult to fully share. Therefore, the distributed scheme designed in this invention is suitable for anti-interference problems in privacy scenarios, making efficient channel and time slot allocation decisions using only local observations, without requiring global information sharing or obtaining private information.

[0041] 4. This invention designs a multi-quota stable matching and local repair mechanism, performing rematching only on a small number of nodes affected by interference or cascading effects. Existing time slot reallocation methods frequently adjust the scheduling of all nodes in pursuit of globally optimal energy efficiency, resulting in high complexity and unsuitability for dynamically coexisting volume area network scenarios. This invention only makes local adjustments, thereby reducing the scale of reconfiguration while maintaining system scheduling stability and robustness. Furthermore, the time slot allocation not only considers interference issues but also takes into account the current buffer accumulation status of each sensor node, which can reduce overall latency. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a topology diagram of the body area network in an embodiment of the present invention.

[0044] Figure 2 This is a schematic diagram of inter-network interference in a coexisting domain network according to an embodiment of the present invention.

[0045] Figure 3 This is a superframe structure diagram of communication between nodes and the coordinator in an embodiment of the present invention.

[0046] Figure 4 This is a flowchart of the heterogeneous area network interference management method based on interference perception prediction and time slot allocation according to the present invention.

[0047] Figure 5 This is a flowchart illustrating the predicted interference power values ​​for each time slot in the channel required for time slot allocation in this invention.

[0048] Figure 6 This is a schematic diagram of all candidate time slots that meet the time delay constraints based on the time dimension of this invention.

[0049] Figure 7 This is a schematic diagram illustrating the classification of sensor nodes based on the temporal continuity characteristics of the allocated time slots of the nodes, as presented in this invention.

[0050] Figure 8 This is a comparison chart of the average packet loss rate of nodes under different numbers of coexisting body area networks for three body area network interference management methods.

[0051] Figure 9 This is a comparison chart of the average packet loss rate of nodes under different numbers of channels for three body area network interference management methods.

[0052] Figure 10 This is a comparison chart of the average node energy efficiency of three body area network interference management methods under different numbers of coexisting body area networks.

[0053] Figure 11 A comparison of the average node energy efficiency of three body area network interference management methods under different numbers of channels. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] The purpose of this invention is to provide a heterogeneous area network interference management method and system based on interference perception prediction and time slot allocation, which aims to solve or improve at least one of the above-mentioned technical problems.

[0056] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0057] Example 1:

[0058] This embodiment provides a heterogeneous body area network (BAN) interference management method based on interference sensing prediction and time slot allocation. The BAN consists of a coordinator and several sensor nodes, which are typically attached to or embedded in the human body, such as... Figure 1As shown; when multiple body area networks (BANs) coexist in close proximity, the signals emitted by some sensor nodes, even after fading, still interfere with the signal reception of the coordinators of adjacent BANs, leading to a decrease in BAN performance. Inter-BAN interference, such as... Figure 2 As shown; the communication between the coordinator and the sensor nodes is based on superframes as the time unit. A superframe consists of the Algorithm Execution Phase (ARP), the Data Transmission Phase (MAP), and the Interference Detection Phase (IDP). The structure of a superframe is as follows: Figure 3 As shown; during the superframe IDP phase, the coordinator can switch to each channel to sample, obtain the interference noise of each channel in the current time slot, and use it as observation data and historical data to predict the interference situation of the channels in future superframes; wherein, the IDP is in the time slots when the sensor nodes are silent and dormant, and the sensor nodes of the body area network do not need to participate in communication; the overall process of the resource allocation method described in this embodiment is as follows Figure 4 As shown, the method includes the following steps:

[0059] S1, Coordinator on the current channel Interference observation is performed, and channel transition sampling is conducted during periods when no data is received to update the interference weights and node propagation factors of time slots. The interference weights of time slots are used to characterize the interference situation of externally coexisting body area networks within the superframe, so that channel-level interference prediction can be reasonably refined into equivalent interference power at the time slot level. The node propagation factors are used to characterize the long-term propagation differences between nodes caused by human posture and tissue occlusion within the body area network, thereby supporting subsequent time slot allocation and local scheduling optimization.

[0060] S1.1 In a body area network, the coordinator observes the channel in each time slot where a sensor node is transmitting data, i.e., during the data transmission phase (MAP) reception process, it observes the interference noise power in real time. For example, on a CC2420 or IEEE 802.15.6 compatible chip, the RSSI register is used to record the total power in the channel, including signal power, interference power, and noise floor. At the same time, during demodulation, the chip calculates the link quality indicator based on the demodulation quality (such as symbol spacing) and maps the current signal-to-interference-plus-noise ratio (SINR) to the link quality indicator. Since the receiver noise floor is usually a relatively fixed reference value after the hardware design is completed, the instantaneous interference power can be separated by subtracting this constant term.

[0061] S1.2 In a body area network, the coordinator switches to other channels during time slots and sleep periods (IDP) when no sensor nodes are transmitting data, collects the energy of that channel, and obtains the interference noise power of the current channel. Within each time slot, the coordinator can sequentially switch to multiple channels for monitoring;

[0062] S1.3, The coordinator records each channel In frame Time slot Perceived interference noise In the sliding window The sum of interference and noise received in this time slot is calculated within the frame and divided by the mean interference of all time slots in the same channel:

[0063]

[0064] in, This represents the total number of time slots within a single frame. For frames within a sliding window, This refers to a time slot within a single frame.

[0065] If certain time slots A value greater than 1 indicates that the interference in this time slot is high relative to the average interference, and vice versa. By performing exponentially weighted recursion, the channel is obtained. In frame Time slot Time slot interference weight:

[0066]

[0067] in, For slot interference weights The time smoothing factor is set to 0.55 in this embodiment. This is the time slot interference weight obtained from the previous time slot.

[0068] S1.4, Coordinator in sliding window Intra-frame statistics of data packet transmission success rate at each node And calculate the moving average; for sensor nodes In the frame The received power after taking the moving average is

[0069]

[0070] Divide it by the average received power of the nodes in the network to obtain the node's power. In frame Transmission factors:

[0071]

[0072] in, This represents the total number of sensor nodes within the body area network. Similarly, the obtained node propagation factors are recursively weighted exponentially to enhance the ability to track recent changes:

[0073]

[0074] in, For node propagation factors The time smoothing factor is set to 0.45 in this embodiment. This is the node propagation factor obtained in the previous time slot.

[0075] S2. The body area network coordinator uses an Interacting Multiple Model (IMM) to predict the channel-level average interference intensity in the next few frames. Each model is updated and predicted using a Kalman filter. Finally, the packet loss probability of each channel in the next few frames is obtained based on the channel-level average interference intensity.

[0076] S2.1 The body area network coordinator obtains the interference power of each channel's partial time slot by observing the interference of the currently used channel and sampling the channel transitions during the sleep period; the coordinator is defined in the first... Frame slot At that time, the channel is obtained The interference noise is In the channel ,frame The set of sampling time slots within is Then the body area network senses the channel. ,frame The average interference intensity is:

[0077]

[0078] S2.2, The body area network coordinator models the channel interference state for each channel. Define its position in the frame. The disturbance state vector is

[0079]

[0080] in, For the first Frame in channel The average interference intensity perceived above, It is the rate of change of interference power over time, used to characterize the rising or decaying trend. The introduction of this extended state structure aims to improve the stability of multi-step prediction, preventing the predicted trajectory from degenerating into a random walk; at the same time, discrete mode variables are introduced. The interference evolution mechanism is represented by three modes: stable interference, moderately varying interference, and sudden interference. This system constitutes a switched linear dynamic system (SLDS). The characteristics of the three modes are shown in Table 1.

[0081] Interference sub-mode Stable disturbance, slow change medium rate of change Sudden interference Typical scenarios The human body is stationary or moving slowly, there are few WBANs around it, and the relative position changes slowly; Users are clustered in a semi-crowded area, their bodies make periodic movements (walking, arm swinging), and their relative distances change, but there is no severe obstruction; A sudden approach of an unfamiliar WBAN, a person suddenly turning around, a sudden LOS / NLOS switch; Physical performance The interference power is low, fluctuates slowly, and packet loss is occasional. The disturbance level fluctuates but is bounded, and there is a trend of "mean decline"; The interference power suddenly changes, and the duration is short or moderate.

[0082] S2.3, The body area network coordinator creates the state transition model for each mode, in the mode Below, channel State evolution can be written as:

[0083]

[0084] in, It is process noise; It is a pattern The state transition matrix; , is the state transition matrix of mode 1, indicating that this mode is approximately a random walk with a small rate of change; , is the state transition matrix for mode 2. The memory length for controlling the rate of change; in this embodiment, the parameter The value is determined by the smoothed interference sequence. Find the difference, and then fit it with a first-order autoregressive model (AR (1)). Based on the fitting results of the AR (1) model, solve it by maximum likelihood estimation. , is the state transition matrix for mode 3, here By designing large process noise allow A jump occurs;

[0085] S2.4. The coordinator of the body area network creates an observation model. In the system, the coordinator cannot directly measure the first derivative of the interference power. Therefore, this variable is modeled as a hidden state, representing the average interference power observed over time series. Indirect estimation; defining the observation model:

[0086]

[0087] in, , It is observation noise; It does not appear in the observation equation, but is represented by the state transition matrix. The coupling terms in the model are incorporated into the prediction step; for example, in a trend model (Mode 2). This means that during time updates:

[0088]

[0089] With this structure, the Kalman filter can utilize consecutive frame observation sequences. Correction At the same time, it will correct in the opposite direction. This makes it a latent variable that can be recursively estimated from the data;

[0090] S2.4, The coordinator of the body area network uses a Markov transition matrix.

[0091]

[0092] Combining the three models, among which Representation pattern Transfer to mode The probability for each channel; The IMM performs the following steps:

[0093] S2.4.1 Interactive Hybridization – The coordinator of the volume area network calculates the initial hybridization conditions for each filter at the current time, given the known mode. No. Frame mode probability State estimation Covariance Calculate the entry mode Mixed probability Initial mixing state and the initial covariance matrix :

[0094]

[0095] S2.4.2, Mode Filtering—The coordinator of the volume area network performs a Kalman filter on each mode; first, state prediction is performed. Based on the mixed initial state of the previous time step, the state at the current time step is predicted through the state transition matrix, providing prior estimates for subsequent measurement updates:

[0096]

[0097]

[0098]

[0099] in, It is a pattern In frame The predicted value at time t is not corrected by the observation at the current time. Simultaneously, covariance prediction is performed, based on the mixed initial covariance of the previous time step, to predict the uncertainty of the current state estimate, taking into account the additional uncertainty introduced by process noise, thus providing prior information for subsequent measurement updates.

[0100]

[0101] in, It is a pattern In frame The prior state covariance reflects the uncertainty of prior state estimation. It is a pattern The process noise covariance matrix is ​​then used; the prior estimate is corrected using the current observations to obtain a more accurate posterior state and covariance; mode In frame The update of the time-kalman gain is given by the following formula:

[0102]

[0103] in, For frames Observational noise covariance at time; update posterior state:

[0104]

[0105] in, It is a pattern In the The posterior state estimate at frame time, i.e., the final state after observation; updating the posterior covariance:

[0106]

[0107] in, It is a pattern In the posterior state covariance at frame time It is the identity matrix;

[0108] S2.4.3 Likelihood and Model Probability Update—The coordinator of the volume area network quantifies the degree of matching between the observed data and the prediction results of each model, dynamically updates the confidence of each candidate model, and provides a weighting basis for subsequent interactive mixing steps; first, the deviation between the observed values ​​and the model's prior predictions is calculated, and the model... In the Frame-time observation residuals It is given by the following formula:

[0109]

[0110] To assess the uncertainty of the residuals, a calculation model is used. In the Residual covariance at frame time :

[0111]

[0112] Based on the residuals and their covariance, a Gaussian likelihood function is constructed to measure the fit between the model and the observations. The calculation model... In the Likelihood value per frame :

[0113]

[0114] in With zero mean and covariance The Gaussian probability density function; the higher the likelihood value, the closer the current observation is to the model. The closer the prediction results are to the model's predictions, the higher the model's confidence level. Finally, by combining the likelihood value and the Markov transition probability, the global probability of each model is updated. In the The global probability at frame time is:

[0115]

[0116] S2.4.4, Fusion Output—Using the global probabilities of each model as weights, a weighted average is performed on the posterior state estimates of all modes to obtain the... Global state estimation of frames :

[0117]

[0118] S2.5, Coordinator Computation in Body Area Networks Frame interference provides a basis for channel selection; based on the currently updated state and probability, the first... Global state estimation of frames for:

[0119]

[0120] Take the first component of this vector as the interference prediction value of the game input. ;

[0121] S2.6 The coordinator in the body area network obtains the packet loss probability of each channel in the next few frames based on the interference power estimation results output by the IMM; for channels operating in the body area network... In a body area network, the SINR of the coordinator in a frame can be expressed as:

[0122]

[0123] in, The power consumption for sending data by sensor nodes. The average path gain of the link. The background noise power of the channel; in this embodiment, BPSK coding is used, and the packet loss rate is expressed as:

[0124]

[0125] in, The length of the data packet in bits. For the current frame, For predicting frames, Let be the signal-to-interference-plus-noise ratio (SIR) of the body area network in channel c and the (t+τ)th frame. It is a complementary error function, defined as:

[0126]

[0127] S3. The coordinator creates a utility function for a non-cooperative game, considering the energy efficiency and interference penalty of switching each channel, and selects a channel with the corresponding probability. Channel selection is performed every few frames (e.g., 10 frames) in the body area network. If interference occurs, time slot matching can be performed again after each frame without reselecting a channel. If adjusting the time slot still cannot alleviate the interference, the utility function calculation and channel selection are triggered in advance. In the body area network, for privacy and security reasons, information exchange between each body area network should be minimized. Each body area network user is rational and self-interested (meaning that a single body area network always uses the strategy that maximizes its own benefits as the optimal strategy) and makes decisions independently.

[0128] S3.1 Establish an inter-network game model for wireless body area networks, allowing players to... ,yes These are independent wireless body area networks that compete for channels, forming a game; players use information they perceive to infer the actions of other players from a set of available actions. In this scheme, the action performed is to select a channel and define the body area network. As one of the players The action performed is In this game theory model, the set of actions of other players is: Players Select Action The obtained utility function is defined as ;

[0129] S3.2 The body area network coordinator creates a utility function to evaluate the benefit gained from each action. Since the actions of other players are unknown, this utility function is set by the coordinator in the [number]th [stage / phase]. The specific steps for frame awareness and profit prediction are as follows:

[0130] S3.2.1, Body Area Network The coordinator calculation is based on channel average interference power prediction and action selection. Expected throughput at time

[0131]

[0132] in, It is the number of predicted frames. It is a body area network No. The number of time slots required for a frame It's the data rate. The length of each time slot, For Body Area Network No. Packet reception rate of the frame;

[0133] S3.2.2, Body Area Network The coordinator calculates the energy consumption of all sensor nodes over future frames, including the energy consumption for transmitting data, sleeping, and switching channels:

[0134]

[0135] in, This refers to the power consumption of the sensor node during sleep mode. This represents the power consumption per time slot when the sensor node is in the data transmission state. For Body Area Network The number of sensor nodes in the middle, Energy consumption for switching a sensor node through a channel;

[0136] S3.2.3, Body Area Network The coordinator calculates the benefit of each action based on the utility function:

[0137]

[0138] in, The throughput obtained in step S3.2.1 The energy consumption obtained in step S3.2.2;

[0139] S3.3, Body Area Network Coordinator Statistics Past The actual throughput of the frames; if the total throughput does not reach the predetermined threshold, then the body area network... The coordinator, based on the payoff of an action, uses a Boltzmann selection strategy to generate an action probability distribution. From the set of actions whose utility function is greater than the previous action, it selects an action with the following probabilities:

[0140]

[0141] in, Temperature is a parameter used to adjust the system's balance between exploration and utilization. This serves as an index for the available channels. This mechanism allows nodes to prioritize efficient solutions while still retaining a certain probability of trying other feasible allocations, thus avoiding getting trapped in local optima and preventing all players from simultaneously jumping to the same "best" channel; if the total throughput reaches a set threshold, the original channel is maintained.

[0142] S3.4 The coordinator in the body area network determines whether to re-interact prematurely based on the packet loss rate of each sensor node. When most sensor nodes are continuously interfered with and local time slot adjustments cannot alleviate the problem, the system will trigger channel selection in advance, re-selecting the channel according to the utility function (e.g., originally triggered every 10 frames) to avoid the continuously deteriorating interference environment. Specifically, when the number of affected sensor nodes that meet the packet loss rate threshold... Reaching the lower limit At this time, channel switching is triggered:

[0143]

[0144] in, The number of frames since the last channel selection in the current frame. For nodes In the past Packet loss rate within a frame (value range from 0 to 1, where 0 indicates no packet loss and 1 indicates that all data packets in the frame are lost). The preset packet loss rate threshold; if the sensor node in the past The average packet loss rate within each frame is greater than Determine sensor nodes It was affected by interference;

[0145] S4. Based on local observations, IMM output, and the interference weight of the time slot, calculate the predicted interference power values ​​for each time slot of the channel required for time slot allocation. The calculation process is described in [reference needed]. Figure 5 :

[0146] S4.1 For each frame, the body area network coordinator determines whether it is the first frame after the game selection; if it is not the first frame, since the coordinator has high-density historical observation data of the channel, the interference power of each time slot of the previous frame is directly used as the basis for subsequent time slot allocation.

[0147] S4.2 If the current frame is the first frame after the game selection, but the game selection action is to use the original channel, then the high-density historical observation data of the previous frame will be used as the basis for subsequent time slot allocation.

[0148] S4.3 If the current frame is the first frame after the game selection, but the game selection action is to switch to a new channel, since the previous sampling of other channels was sparse and lacked time slot level resolution; therefore, the predicted average interference power of this channel at future times, output by the IMM filter, is used. Interference weights with time slots Multiplying these together yields the predicted interference power for each time slot in the next frame:

[0149]

[0150] S5. The body area network coordinator calculates the sensor node's preference for time slots and the time slot's preference ranking for sensor nodes based on the predicted interference power of each time slot and the historical path loss obtained from sliding window statistics. It then uses bilateral matching to allocate time slots to each body area network. If interference occurs, local adjustments are made, specifically including the following steps:

[0151] S5.1 The coordinator calculates the time slot quota required by each sensor node in the body area network; defines the body area network. Sensor nodes in The amount of data to be uploaded is Then body area network Medium sensor node The required time slot quota is:

[0152]

[0153] in, It's the data rate. The length of each time slot;

[0154] S5.2 The coordinator calculates the preference ranking of each sensor node in the volume area network for each time slot within the superframe. The specific steps are as follows:

[0155] S5.2.1 At the beginning of each scheduling cycle, the body area network coordinator first calculates the remaining tolerable delay window for each sensor node, thereby constraining the range of selectable time slots and reducing the search space, such as... Figure 6 As shown; in addition, all candidate time slots that meet the time delay constraints are screened from the time dimension to avoid data timeout and discarding due to scheduling delays, ensuring that the scheduling scheme meets the hard latency requirements of the nodes; sensor nodes are defined. The remaining tolerable delay is:

[0156]

[0157] in, For sensor nodes The maximum tolerable latency is determined by the type of service (for example, the maximum latency of an ECG monitoring node is usually set to 200 ms, while the maximum latency of a motion posture sensor node can be relaxed to 500 ms). The waiting time up to the current frame reflects the backlog of nodes; therefore, the set of available time slots is:

[0158]

[0159] in, Let n be the set of optional time slots for sensor node n. This represents the total number of time slots within a single frame.

[0160] S5.2.2. Sensor nodes preferentially select time slots that can meet the target signal-to-interference-plus-noise ratio (SINR). Based on the interference situation prediction results, the set of selectable time slots is... Perform hierarchical sorting; first, calculate each candidate time slot. Predicted SINR value:

[0161]

[0162] in For nodes Historical path loss obtained from sliding window statistics The power consumption for sending data by sensor nodes. The background noise power of the channel. For Body Area Network Inner candidate slot In the Predicted interference power for each frame; For each candidate time slot The predicted SINR values ​​are divided into three categories: 1. Time slots that meet the requirements: Predicted SINR ≥ node target SINR threshold. These time slots ensure reliable data transmission and are listed as the first priority; for high-priority nodes, the judgment threshold is set even higher. 2. Time slots that exceed the requirements: Predicted SINR is significantly higher than the target threshold (e.g., exceeding by more than 3dB). These time slots have high communication quality redundancy and are listed as the second priority. 3. Time slots that do not meet the requirements: Predicted SINR < node target SINR threshold. These time slots are only used as alternatives when resources are scarce and are listed as the third priority. Time slots in each category are arranged in descending order of SINR value, and then the rankings of the three categories are merged. Through this hierarchical ranking, the system prioritizes allocating time slots that meet the communication quality standards to nodes. Under the premise of meeting basic transmission requirements, it makes the most of redundant resources to improve transmission stability, while avoiding wasting time slot resources due to excessive pursuit of high redundancy.

[0163] S5.2.1 For high-priority nodes, time slots with similar signal-to-noise ratios are sorted in ascending order of time proximity to ensure that data from high-priority nodes is transmitted in the time slot closest to its generation time. For high-priority nodes such as medical monitoring nodes, the latency sensitivity is much higher than that of ordinary nodes. Therefore, in time slots with similar signal-to-interference-plus-noise ratios, a second sorting based on time proximity is introduced to minimize data transmission delay. Define sensor nodes. In the time slot The time proximity is:

[0164]

[0165] in, For time slots absolute timestamp, For sensor nodes The closer the time slot is to the data generation time, the smaller the transmission delay.

[0166] S5.3 The coordinator calculates the preference ranking of each time slot for all sensor nodes within the superframe. The specific steps are as follows:

[0167] S5.3.1 At the beginning of each scheduling cycle, based on the temporal continuity characteristics of the time slots allocated to the nodes, all sensor nodes are divided into two categories; the first category is those that can be allocated time slots. Sensor sets merged into continuous time slots The allocated time slots of such nodes can be merged with the currently unallocated time slots into a continuous time interval; merging continuous time slots can reduce the frequent start-up and shutdown of node radio frequency modules, thereby reducing the node's power consumption; such as Figure 7 As shown, the allocated time slots of nodes 3 and 5 are temporally adjacent to the current candidate time slot, and can be merged into a continuous transmission window, thereby reducing the number of active and dormant handovers. The second category is sensor sets whose continuous time slots cannot be merged. The allocated time slots of such nodes cannot form a continuous interval with the currently unallocated time slots; the core purpose of this classification strategy is to maximize the utilization of continuous time slots and reduce the overall energy consumption of the system by reducing the number of node wake-ups.

[0168] S5.3.2, According to dynamic utility value To quantify the scheduling priority of nodes, various sensors are sorted in ascending order. A dynamic utility function is defined for each node, which integrates latency urgency and business priority, and can dynamically reflect the scheduling needs of the node.

[0169]

[0170] in, For nodes Remaining tolerable latency, For nodes Effective priority, and node Basic priority Node propagation factors related:

[0171]

[0172] The basic priority of a node is related to its business and function. For example, medical monitoring nodes are set to high priority, while ordinary wearable nodes are set to low priority. The node propagation factor is obtained by the coordinator by counting the received power of each node's data packets within a sliding time window and performing a sliding average. It is used to characterize the differences in channel propagation loss within the body area network caused by factors such as changes in human posture and tissue occlusion.

[0173] S5.3.3, Sort the sensor set that can be merged into consecutive time slots. With sensor sets that cannot be merged into continuous time slots Merge to generate time slots Preference ranking for each sensor node:

[0174]

[0175] S5.4 For sensor nodes that have not met their quotas, a matching request is sent to the highest-ranked time slot in their preference list that has not yet rejected them. This process ensures that nodes always prioritize trying to match the time slot that best meets their needs, improving matching efficiency.

[0176] S5.5 After receiving a matching request, the time slot compares the currently temporarily matched sensor (if any) with the new requester, selects the better one according to its own preference list. If the new requester is better, the original matcher is rejected and the time slot is allocated to the new requester; if the original matcher is better, the new requester is rejected directly and the original match is maintained.

[0177] S5.6 If a rejected node still does not meet its quota requirements, it will continue to propose to the next time slot in its preference list, looping until all sensors meet their quotas, or no more time slots can be found to propose to. This iterative process ensures that the system converges to a stable matching state within a finite number of steps.

[0178] S5.7 When an event is detected at a certain sensor node, the following local repair process is initiated in the next frame to avoid the high overhead of global rematching. The specific steps are as follows:

[0179] S5.7.1 Freeze Matching: Keep the matching relationship between undisturbed nodes and time slots unchanged, and maintain their original preference order to ensure communication continuity.

[0180] S5.7.2, Unmatch: Only unmatch the interference of nodes. According to steps S5.2 and S5.3, update the preference order of the interference nodes for time slots and the preference order of time slots for interference nodes.

[0181] S5.7.3 Re-initiating the request: The disturbed node initiates a new matching request to the optimal time slot based on the updated preference list.

[0182] S5.7.4, Local comparison on the time slot side: The time slot only performs a local comparison between the current occupant and the new requester, and selects the node with better utility to complete the matching, without having to traverse all candidate nodes.

[0183] Example 2

[0184] Referring to steps S1 to S5 of the heterogeneous body area network interference management strategy based on interference perception prediction and time slot allocation disclosed in Example 1, this example uses MATLAB to simulate and evaluate the performance of the proposed scheme. The specific operation method of network operation is as shown in Example 1. Based on this, this example tests the change of the average packet loss rate of sensor nodes with the number of coexisting body area networks and channels when using this scheme and two other schemes. Some important parameters used in the simulation in this example are as follows: the number of sensor nodes in the body area network is between 5 and 20, the superframe length is 40 ms, the time slot length is 2 ms, the data rate is 200 kbps, the data packet size is 25 bytes, and the selected noise is -114 dBm. The movement of the body area network adopts a Gauss-Markov movement model with an average speed of 1 m / s and a random factor of 0.3. All simulation results are obtained by averaging multiple independent simulated motion simulations. Figure 8 and Figure 9 The comparisons are as follows: our proposed scheme (IMM-based Prediction and Channel Allocation, IPCA), CORS proposed by Huang et al., and DOM proposed by Guo et al., under different numbers of coexisting domain networks and different numbers of idle channels, showing the average packet loss rate of nodes.

[0185] Figure 8The graph describes how the average packet loss rate of a single node changes with the increase in the number of coexisting body area networks (BANs). As the graph clearly shows, with the increase in the number of coexisting BANs, the congestion between BANs increases, limited channel and time slot resources are shared by more nodes, and co-frequency and co-slot transmissions result in stronger interference for each BAN, significantly increasing the collision probability and the average packet loss rate. Based on this, the proposed IPCA consistently achieves the lowest packet loss rate across different node scales. Its key lies in IPCA's accurate prediction of stable, slowly varying, and bursty interference using an Interactive Multiple Model (IMM), combined with non-cooperative game theory to select channels with higher returns, thus avoiding interference at the source and efficiently repairing local conflicts. In contrast, CORS lacks proactive interference prediction capabilities and relies on fixed Latin square packet allocation. In high-density resource competition scenarios caused by a surge in the number of BANs, the flexibility of fixed packet rules is insufficient. Furthermore, the increase in the total number of retransmission nodes also leads to an increased probability of hash collisions, and secondary interference further pushes up the packet loss rate. While the DOM scheme achieves interference identification through a Naive Bayes classifier and uses the Hungarian algorithm for time slot allocation to pursue optimal global energy efficiency, its centralized decision-making method makes it insufficiently adaptable to dynamically changing interference environments. Furthermore, in order to ensure that each sensor node occupies only one time slot to complete transmission, the scheme compresses transmission time by increasing the transmission rate of large data nodes. Under the same interference level, a higher transmission rate places more stringent requirements on channel quality and is prone to data packet demodulation failure due to insufficient signal-to-noise ratio, thereby indirectly increasing the probability of packet loss.

[0186] Figure 9 The graph describes how the average packet loss rate of a single node changes with the increase in the number of channels. As can be seen from the graph, the average packet loss rate of all three schemes decreases significantly with the increase in the number of available channels. The core reason is that more channel resources alleviate resource competition among coexisting body area networks (BANs), reducing collisions in co-channel transmissions and thus reducing packet loss caused by interference. The CORS scheme consistently maintains the highest packet loss rate; even with an increase in the number of channels, its packet loss rate decrease is still weaker than other schemes. This is because CORS relies on fixed Latin square packet allocation and lacks dynamic adaptation capabilities to channel resources. Even with an increase in channels, the fixed rules still limit resource utilization efficiency. The DOM scheme achieves globally optimal time slot allocation using the Hungarian algorithm, but when facing strong interference scenarios that cannot be mitigated by time slot adjustments, its channel switching strategy lacks targeted interference perception and decision-making basis, resulting in a slightly higher packet loss rate than IPCA. IPCA consistently maintains the lowest packet loss rate, and its packet loss rate decreases most significantly with the increase in the number of channels. This is due to IPCA's interference prediction and channel selection mechanism, which allows the BAN to autonomously select low-interference channels, maximizing the utilization of newly added channel resources and effectively avoiding interference.

[0187] Example 3

[0188] Referring to steps S1 to S5 of the heterogeneous area network interference management method based on interference sensing prediction and time slot allocation disclosed in Example 1, this example uses MATLAB to simulate and evaluate the performance of the proposed scheme. The specific operation method of network operation is as shown in Example 1. Based on this, this example tests the change in the average energy efficiency of sensor nodes with the number of coexisting area networks and channels when using this scheme and two other schemes. The calculation methods of some important parameters and results used in the simulation in this example are consistent with those in Example 2. Figure 10 and Figure 11 The comparisons are as follows: the average energy efficiency of the present scheme (IMM-based Prediction and Channel Allocation, IPCA) with that of CORS proposed by Huang et al. and DOM proposed by Guo et al. under different numbers of coexisting domain networks and different numbers of idle channels.

[0189] Figure 10 This describes how the average energy consumption of a single node changes with the increase in the number of coexisting body area networks (BANs). The increase in the number of BANs intensifies resource competition and inter-network interference. Increased packet loss leads to a higher retransmission rate, requiring nodes to consume more energy for data retransmission and channel switching, thus increasing the total energy consumption in the network and consequently increasing the average power consumption of a single node. The IPCA scheme maintains a relatively high level of energy efficiency. Its core lies in the accurate prediction of dynamic interference by the Interactive Multiple Model (IMM) and the efficient utilization of low-interference resources by non-cooperative game-theoretic channel selection, reducing the additional energy overhead of nodes due to retransmissions caused by interference. The DOM scheme's energy efficiency falls between IPCA and CORS. Although DOM can achieve global optimum on the time slot side through the Hungarian algorithm, its centralized decision-making and the high complexity of the time slot allocation algorithm during interference make it insufficiently adaptable to high-density interference, and some nodes still need to bear a significant amount of retransmission energy consumption. Furthermore, DOM compresses transmission time by increasing the node's transmission rate, ensuring that a single node occupies only one time slot. However, according to Shannon's formula, transmission power and rate are non-linearly positively correlated. When transmitting the same amount of data, increasing power to compress transmission time actually requires more energy, further increasing node energy consumption. The CORS scheme consistently maintains the lowest energy efficiency because it lacks an active interference prediction mechanism and relies solely on fixed Latin square packets for resource allocation. In high-density scenarios with a surge in the number of body area networks, the fixed packet rules cannot flexibly cope with the intensified resource competition, leading to a large number of nodes entering the retransmission phase due to transmission conflicts. At the same time, although CORS uses a large modulus hash function to differentiate retransmission time slot allocation, the surge in the total number of retransmission nodes still triggers secondary interference. Nodes need to continuously consume energy for retransmission and channel switching, ultimately resulting in a significant increase in energy loss.

[0190] Figure 11The graph describes how the average energy consumption of a single node changes with the increase in the number of channels. As can be seen from the graph, the average energy consumption of all three schemes increases significantly with the increase in the number of available channels. This is because the increase in the number of channels directly expands the resource selection space, reducing retransmissions and power waste caused by interference. Therefore, the energy efficiency of each scheme improves with the increase in the number of channels. IPCA consistently maintains the highest energy efficiency, and its energy efficiency increase is the most significant with the increase in the number of channels. This is because IPCA performs interference prediction and channel and time slot coordinated scheduling. More channels provide it with a richer selection of low-interference resources. Combining game theory to select channels with better energy efficiency and local time slot allocation, nodes do not need to frequently adjust power or retransmit, effectively reducing energy loss. CORS relies on fixed Latin square group allocation of resources, resulting in insufficient utilization efficiency of newly added channels, and the passive retransmission mechanism increases energy consumption, leading to the smallest improvement in energy efficiency.

[0191] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0192] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A heterogeneous area network interference management method based on interference sensing prediction and time slot allocation, characterized in that, include: S1. The coordinator observes the interference power of the current channel in real time during the data transmission phase, and performs sparse sampling through channel transitions during inactive periods to update the time slot interference weights and node propagation factors of each channel. S2. Establish an interference prediction model to predict the average interference intensity of each channel in the next few frames. S3. Construct an inter-network game model, treating each heterogeneous body area network as a game participant, evaluating the benefits of each channel through prediction results, conducting non-cooperative games among multiple coexisting body area networks, and selecting the communication channel that maximizes the benefit function of the inter-network game model. Specifically, S3 is: S3.1 Treat each heterogeneous wireless body area network as an independent game participant, construct an inter-network non-cooperative game model, define a utility function with channel throughput and transmission energy consumption as the core, calculate the utility function value of each channel based on the signal-to-interference-plus-noise ratio and packet loss probability obtained from interference prediction, and use it as the payoff value of each channel. S3.2 Statistically analyze the actual throughput of historical communications. If the preset throughput threshold is not reached, use the Boltzmann selection strategy to generate the selection probability distribution of each channel and select the communication channel according to the probability. If the threshold is reached, continue to use the original communication channel. S3.3 Real-time statistics of packet loss rate of each sensor node. When the number of interfered nodes that meet the preset packet loss rate threshold reaches the set lower limit, the re-game process is immediately triggered to re-execute the channel benefit assessment and non-cooperative game and select a new communication channel. S4. Based on the prediction results of the interference prediction model, local observations, and time slot interference weights, obtain the predicted interference power value for each time slot in the channel. S5. Based on the predicted value of time slot interference power and the node propagation factor, calculate the sensor node's preference ranking for time slots and the time slot's preference ranking for sensor nodes, respectively, allocate time slots to sensor nodes, and perform local time slot rematching operation on the interfered node when interference is detected. Specifically, S5 is: S5.1 Calculate the required time slot quota for each sensor node based on the amount of data to be uploaded, the data transmission rate, and the fixed length of the time slot for each sensor node. S5.

2. Based on the remaining tolerable delay of sensor nodes, candidate time slots are screened, and the preference ranking of each sensor node for time slots is generated by layering according to the signal-to-interference-plus-noise ratio and combining the time proximity. S5.3 Classify the nodes according to the temporal continuity characteristics of the time slots allocated to the sensor nodes, and generate a preference ranking of the sensor nodes for each time slot by combining the service priority of the fusion node and the dynamic utility value of the node propagation factor. S5.

4. Based on bidirectional preference sorting, the initial time slot allocation is completed. Nodes that do not meet the time slot quota initiate a matching request to the best time slot in their preference list. The time slot selects a better node according to its own preference list to complete the matching update. The rejected nodes initiate requests to the next preferred time slot in turn until all nodes meet the quota or there are no available time slots. S5.

5. Monitor the communication status of sensor nodes in real time and perform local time slot rematching operation on the interfered nodes.

2. The heterogeneous area network interference management method based on interference sensing prediction and time slot allocation according to claim 1, characterized in that, Specifically, S1 is: S1.1 The coordinator observes the current working channel in real time during the data transmission phase and collects the interference noise power of the current channel while receiving data from sensor nodes. S1.2 The coordinator switches to other channels during time slots and sleep periods when no sensor nodes are transmitting data, collects channel energy, and obtains the interference noise power of the current channel; S1.3 The coordinator records the interference noise of each channel, calculates the proportion of interference noise in each time slot within a preset sliding window, and performs exponential weighted recursive calculation on the interference noise proportion to obtain the time slot interference weight of each channel in different time slots. S1.4 The coordinator counts the data packet transmission success rate of each sensor node within a preset sliding window, and at the same time collects and performs a sliding average processing on the received power of each node. The node propagation factor is obtained by dividing the received power after the node sliding average by the average received power of all sensor nodes in the network.

3. The heterogeneous area network interference management method based on interference sensing prediction and time slot allocation according to claim 1, characterized in that, Specifically, S2 is: S2.1 Based on the collected multi-channel interference data, calculate the average interference intensity of each channel in each frame, define an interference state vector including average interference intensity and interference power change rate, introduce three interference modes: stable interference, moderately variable interference, and burst interference, and construct a switching linear system. S2.

2. Create a corresponding state transition model and observation model for each interference mode. Fuse the state transition models and observation models of the three interference modes through the Markov transition matrix. Perform interactive mixing, mode filtering, likelihood and mode probability update, and fusion output operations in sequence to complete the global state estimation of interference for each channel. S2.3 Based on global state estimation, predict the average interference intensity of each channel in the next few frames, and derive the signal-to-interference-plus-noise ratio and packet loss probability of each channel in the next few frames based on the average interference intensity.

4. The heterogeneous area network interference management method based on interference sensing prediction and time slot allocation according to claim 1, characterized in that, Specifically, S4 is: First, a game of channel switching is played. Then, the average interference intensity output by the interference prediction model is multiplied by the time slot interference weight of each time slot to obtain the predicted interference power value of each time slot of the channel.

5. The heterogeneous area network interference management method based on interference sensing prediction and time slot allocation according to claim 1, characterized in that, Specifically, S5.5 is as follows: The communication status of sensor nodes is monitored in real time. When interference is detected, the time slot matching relationship of the undisturbed nodes is frozen, and only the matching relationship of the interfered nodes is released and their bidirectional preference ranking is updated. The interfered nodes re-initiate matching requests based on the updated preference list. The time slots are only locally compared and matched between the currently occupying nodes and the newly requesting nodes, thus realizing local time slot rematching of the interfered nodes.

Images