Intelligent garbage bin low-power-consumption communication method based on NB-IoT

Abstract

The invention relates to the technical field of wireless communication networks, in particular to an intelligent garbage bin low-power-consumption communication method based on NB-IoT, and the method comprises the following steps: S1, collecting the garbage capacity, gas concentration, temperature and humidity and uplink failure history of a garbage bin, and constructing a communication attempt record with a timestamp in combination with geographical location information; s2, inputting the communication attempt record into a conflict prediction model based on a time sequence attention mechanism, and outputting a conflict probability of a future communication time slot; s3, dividing power consumption states according to the conflict probability, and setting an asynchronous wake-up window; s4, in combination with a priority ranking result and available resource blocks broadcasted by the NB-IoT network, executing minimum power consumption path matching, and generating a communication triggering plan; and S5, starting the NB-IoT module for data transmission according to the communication triggering plan. According to the invention, by introducing conflict perception, power consumption adaptive scheduling and resource optimal matching mechanisms, efficient and low-power-consumption communication of the intelligent garbage bin under the NB-IoT network is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication network, and particularly relates to a low-power communication method of intelligent fruit peel box based on NB-IoT. BACKGROUND

[0002] With the continuous advancement of smart city construction, intelligent fruit peel boxes, as an important part of fine urban environmental management, are gradually realizing functions such as state perception, remote monitoring and data uploading. NB-IoT (Narrow Band Internet of Things) communication technology has become an important choice for wireless communication of intelligent fruit peel boxes due to its wide coverage, low power consumption, high connection density and other advantages. Through the NB-IoT module, the fruit peel box can report information such as garbage capacity, gas concentration and environmental parameters to the management platform, realizing intelligent management functions such as garbage overflow warning, abnormal detection and operation and maintenance scheduling.

[0003] However, most of the existing communication methods of intelligent fruit peel boxes based on NB-IoT use a fixed period reporting mechanism, lack dynamic perception ability of the communication environment state, and are prone to cause frequent retransmission or delayed reporting during network congestion period, resulting in energy waste and unstable communication. In addition, most schemes do not fully consider the comprehensive influence between access conflict risk, subcarrier resource occupation and terminal power consumption in different communication time slots, and the communication behavior is blind, especially in urban underground space or shielding area, which is more prone to access failure and unreasonable channel resource allocation problems, which seriously restricts the energy efficiency and robustness of the fruit peel box communication system. SUMMARY

[0004] The present application provides a low-power communication method of intelligent fruit peel box based on NB-IoT, which can dynamically identify a low-conflict communication window, control the working mode of the NB-IoT module based on the prediction result, and construct a minimum energy consumption path scheduling plan combined with the communication resource pool, thereby reducing power consumption to the maximum extent while ensuring reliable data upload, improving communication success rate and system stability, and being particularly suitable for intelligent terminal scenarios with complex urban network environment and limited spectrum resources.

[0005] A low-power communication method of intelligent fruit peel box based on NB-IoT, comprising the following steps:

[0006] S1, asynchronous perception and collection: obtaining state data inside the fruit peel box, including garbage capacity, gas concentration, temperature and humidity, and combining the current geographic position of the device and the history of uplink failure to construct a time-stamped asynchronous communication attempt record;

[0007] S2, time slot conflict prediction modeling: inputting the asynchronous communication attempt record into an NB-IoT access conflict prediction model based on a timing attention mechanism to predict the congestion degree of future communication time slots and output the communication conflict probability of the target period;

[0008] S3, adaptive power consumption window division: adjust the power consumption state and access window of the NB-IoT module according to the communication conflict probability, adopt an asynchronous time slice mechanism with a non-fixed period, start the communication module in a predicted low congestion time slot, and set the shortest available communication window;

[0009] S4, time slot optimal matching scheduling: combine the conflict prediction model output and the available subcarrier resource pool provided by the NB-IoT network, perform time slot priority sorting and matching, construct an access plan of the minimum power consumption path, and output an optimized communication trigger plan;

[0010] S5, dynamic execution: according to the communication trigger plan, start the NB-IoT communication module to send target data, and complete the current communication task.

[0011] Optionally, the asynchronous sensing collection in S1 includes:

[0012] S11, synchronous collection of sensing data: use multiple types of low-power sensors arranged inside the fruit peel box to obtain state data, specifically including:

[0013] Garbage capacity data D v : measured by an ultrasonic ranging unit bin .

[0014] Gas concentration C g : measured by a multi-gas sensor s .

[0015] Temperature T s and humidity H s : obtained by an environmental sensor

[0016] S12, device location positioning: call the built-in GNSS module to obtain the geographic location information P lat of the current fruit peel box lon , and timestamp t pos to form a location information vector L pos = (P lat , P lon , t pos );

[0017] S13, uplink failure event statistics and modeling: traverse the device communication log, and in the time window [t i , t j ], count the frequency of failure events in NB-IoT communication attempts, and construct an uplink failure frequency function F fail (t k );

[0018] S14, constructing asynchronous communication attempt record: aggregating the obtained state data, position information vector and uplink failure frequency according to the timestamp t r constructing asynchronous communication attempt record R try (t r ).

[0019] Optionally, the time slot conflict prediction modeling in S2 comprises:

[0020] S21, historical record encoding: linearly encoding the constructed asynchronous communication attempt record sequence {R try (t1),P try (t2),...,R try (t n )} to obtain an input feature sequence matrix X = [x1, x2, …, x n ], wherein each vector x i comprises an encoded representation of the garbage capacity, gas concentration, environmental parameters, position information and failure frequency at time t i ;

[0021] S22, time sequence attention weight calculation: based on a self-attention mechanism, time sequence dependence modeling is performed on the input feature sequence, and an attention weight matrix A between time points is calculated;

[0022] S23, conflict probability prediction output: weighted summation is performed on the value vector sequence V = XW v using the attention weight matrix A as the weight to obtain a context aggregation representation, and a communication conflict probability is output using a fully connected prediction layer for a predicted target time point t n+1 ,t n+2 ,…,t n+m .

[0023] Optionally, the adaptive power consumption window division in S3 comprises:

[0024] S31, power consumption state selection based on conflict probability: receiving the output communication conflict probability sequence, dividing the congestion level of the current period according to the preset congestion classification threshold, including low congestion, medium congestion or high congestion, and according to the congestion level, the NB-IoT module of the intelligent litter bin will switch to the corresponding power consumption working state;

[0025] S32, dynamic asynchronous communication window division: combining the local minimum points in the communication conflict probability sequence, i.e. the predicted low conflict communication empty window period, a sliding window mechanism is adopted to determine the next communication module wake-up time period, each wake-up time period is defined as a dynamic asynchronous communication window [t start ,t end ], and the length is adjusted to be the shortest according to the current power consumption state and the uplink data volume.

[0026] Optionally, the power consumption state selection based on the conflict probability in S31 comprises:

[0027] S311, communication conflict probability receiving: receiving the output future time period communication conflict probability sequence Wherein, m is the number of predicted future time points, P conflict (τ1), P conflict (τ2),..., P conflict (τ m ) are NB-IoT access conflict probabilities at predicted time points τ1, τ2,..., τ m

[0028] S312, congestion level division: set the communication congestion probability grading threshold set Θ={θ L , θ M}, wherein θ L is the lower limit of congestion threshold, θ M is the upper limit of congestion threshold, according to the current conflict probability P conflict (τ cur ) and the congestion grading threshold, the congestion level is divided, specifically including:

[0029] If P conflict (τ cur )≤θ L , it is determined as low congestion;

[0030] If θ L <P conflict (τ cur )≤θ M , it is determined as medium congestion;

[0031] If P conflict (τ cur )>θ M , it is determined as high congestion;

[0032] S313, power consumption state switching execution: according to the current congestion level to control the power consumption state of NB-IoT communication module, specifically including:

[0033] If the current congestion level is low congestion, enter the light wake-up listening state (short interval polling + fast access);

[0034] If the current congestion level is medium congestion, maintain the periodic wake-up state (fixed interval polling);

[0035] If the current congestion level is high congestion, switch to deep sleep state (close the radio module).

[0036] ​Optionally, the dynamic asynchronous communication window division in S32 includes:

[0037] S321, Extraction of local low-conflict time points: Communication conflict probability sequence based on output The neighborhood sliding comparison method is used to extract the time point set that meets the local minimum condition. Expressed as:

[0038]

[0039] Among them, τ k is the kth prediction moment, P conflict (τ k ), P conflict (τ k-1 ), P conflict (τ k+1 ) are respectively the time τ k , τ k-1 , τ k+1 The probability of communication conflict;

[0040] S322, low conflict segment sliding window identification: at each local minimum time point As the center, set the time sliding window radius δ t , define the candidate low-conflict segment as In the window If the maximum conflict probability satisfies It is considered as an effective communication window period, and the window that meets the conditions Included in the set of available communication time slots

[0041] S323, window length minimization adjustment: for each available window According to the current NB-IoT module power consumption status mark λ mode ∈{low congestion, medium congestion, high congestion} and the total amount of data packets to be uploaded S data , calculate the dynamic minimum communication window length Finally, from the collection Select the earliest one to meet The window is used as the next communication wake-up time period [t wake ,t sleep ].

[0042] Optionally, the time slot optimal matching scheduling in S4 includes:

[0043] S41, candidate time slot priority sorting: in the determined available communication time period set In the example, according to the communication conflict probability P of each window center point conflict (τ k), historical communication success rate, and terminal wake-up cost to build a comprehensive priority function, score and sort all available communication time slots, so that the time slots with high predicted success rate and low power consumption are ranked higher;

[0044] S42, resource pool matching and access path construction: Obtain the available physical resource block (PRB) set and subcarrier resource pool information broadcast by the current NB-IoT cell, bind and match high-priority candidate time slots with idle spectrum in the resource pool, consider resource continuity, terminal access rate and scheduling delay constraints during the matching process, and use the minimum power consumption path search strategy to generate the optimal access plan, including the target wake-up time, allocated resource units and transmission duration, and finally form a communication trigger instruction.

[0045] Optionally, the priority sorting of candidate time slots in S41 includes:

[0046] S411, multi-dimensional feature extraction: from the output of the available communication time period set In the above example, the key features of each candidate window are extracted, including the window center point τ k The communication conflict probability p k =P conflict (τ k ), the historical average success rate of uplink communication in the past around this time period (within ±Δ seconds) γ k 、Terminal in window W (k) Energy consumption required to wake up and complete communication k , each window W (k) Corresponding to a feature triple (p k ,γ k ,ε k );

[0047] S412, Priority scoring function construction: define comprehensive priority scoring function Ψ k For each candidate window W (k) Scoring is expressed as:

[0048]

[0049] Among them, ε max ={ε1,...,ε N}, N is the number of available communication windows, α1, α2, α3 are the weighting coefficients of the scoring function;

[0050] S413, priority sorting and output: According to the score value Ψ of each window k right Sort all candidate windows in descending order to form a priority list

[0051] Optionally, the resource pool matching and access path construction in S42 comprises:

[0052] S421, acquiring cell resource information: the terminal device periodically listens to the broadcast channel (NPRS) of the NB-IoT cell where it is located, and acquires the physical resource block set in the current scheduling period and the subcarrier resource pool;

[0053] S422, high priority time slot matching and path scoring: from the output priority list , try to match the physical resource block set in turn, and construct a path scoring function Φ j representing the power consumption cost of each candidate access path, and select the path with the minimum Φ j as the optimal access path

[0054] S423, trigger instruction generation: map the optimal access path to a communication scheduling instruction, including wake-up timestamp, used subcarrier number set, used time slot number set, and expected transmission duration.

[0055] The beneficial effects of the present application are:

[0056] The present application, through asynchronous sensing and collection of the internal state data of the fruit box, geographical position and communication failure history, combines a timing attention mechanism to construct a conflict prediction model, accurately predicts the congestion degree of the future NB-IoT communication time slot, and dynamically adjusts the power consumption state and access window of the communication module according to the conflict probability, significantly improving the adaptability of the system to complex urban weak signal environment, solving the defects of the existing fixed period communication strategy in energy consumption and response delay, and realizing low-power operation and high availability of the fruit box communication system.

[0057] The present application, through local minimum value identification and sliding window mechanism, accurately locates the low-conflict communication window period, and introduces terminal power consumption state and data load sensing to adjust the communication window length, so that the NB-IoT communication module only wakes up and transmits in the time period with high expected success rate and low network load, avoiding the problem of repeated retransmission or invalid waiting in the period of high network congestion, and at the same time, a multi-factor priority scoring function based on conflict probability, historical success rate and wake-up energy consumption is constructed, so that the communication scheduling process not only has network sensing ability, but also takes into account the minimization of terminal power consumption, realizing the dual optimization of scheduling accuracy and energy consumption control.

[0058] The present invention combines the physical resource block set and subcarrier resource pool broadcast by the NB-IoT cell, executes a matching strategy based on time slot priority and resource availability, constructs the optimal access path and generates communication scheduling instructions by minimizing the power consumption path scoring function, effectively improving spectrum resource utilization and access success rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0060] Fig. 1 Schematic diagram of the communication method flow in an embodiment of the present invention;

[0061] Fig. 2 Schematic diagram of time slot conflict prediction modeling according to an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0063] like Figs. 1-2 As shown, a low-power communication method for an intelligent rubbish bin based on NB-IoT includes the following steps:

[0064] S1, asynchronous sensing and collection: This system obtains status data inside the garbage bin, including garbage volume, gas concentration, temperature and humidity. It also combines the device's current geographic location and uplink failure history to create a timestamped asynchronous communication attempt log.

[0065] S2, time slot conflict prediction modeling: asynchronous communication attempt records are input into the NB-IoT access conflict prediction model based on the temporal attention mechanism to predict the congestion level of future communication time slots and output the communication conflict probability in the target period;

[0066] S3, Adaptive Power Consumption Window Division: This adjusts the NB-IoT module's power consumption and access window based on the probability of communication conflicts. It uses a non-fixed-cycle asynchronous time-slicing mechanism to activate the communication module during predicted low-congestion time slots, setting the shortest available communication window.

[0067] S4, time slot optimal matching scheduling: combined with the conflict prediction model output and the available subcarrier resource pool provided by the NB-IoT network, the time slot priority is sorted and matched, the access plan of the minimum power consumption path is constructed, and the optimized communication trigger plan is output;

[0068] S5, dynamic execution: according to the communication trigger plan, the NB-IoT communication module is started to send target data, and the communication task of this round is completed.

[0069] The asynchronous sensing collection in S1 includes:

[0070] S11, synchronous sensing data collection: using multiple types of low-power sensors arranged inside the fruit peel box to obtain state data, specifically including:

[0071] Garbage capacity data D v : the vertical distance d between the garbage surface and the box opening is measured by an ultrasonic ranging unit bin , expressed as:

[0072]

[0073] Where h bin is the internal height of the fruit peel box;

[0074] Gas concentration C g : the concentration of target gas (CH4, NH3) is measured by a multi-gas sensor;

[0075] Temperature T s and humidity H s : the current box environment state is obtained by an environmental sensor;

[0076] S12, device location positioning: call the built-in GNSS module to obtain the geographic location information P lat of the current fruit peel box lon and timestamp t pos , forming a location information vector, L pos = (P lat , P lon , t pos );

[0077] S13, uplink failure event statistics and modeling: traverse the device communication log, and in the time window [t i , t j ], the frequency of failure events in NB-IoT communication attempts is counted, and the uplink failure frequency function F fail (t k ) is constructed, expressed as:

[0078]

[0079] where t k is a discrete time point, N fail is the total number of attempts at time point t k is the number of failures at time point t k is the total number of attempts at time point t total is the total number of attempts at time point t k is the total number of attempts at time point t k is the total number of attempts at time point t

[0080] S14, record construction of asynchronous communication attempts: aggregate the obtained state data, position information vector, and uplink failure frequency according to the time stamp t r to construct the record of asynchronous communication attempts R try (t r ), which is expressed as:

[0081] R try (t r ) = (D v , C g , T s , H s , L pos , F fail (t r )).

[0082] The time slot conflict prediction modeling in S2 includes:

[0083] S21, historical record coding: linearly code the constructed asynchronous communication attempt record sequence {R try (t1), R try (t2),..., R try (t n )} to obtain the input feature sequence matrix X = [x1, x2, …, x n ], where each vector x i includes the coded representation of the garbage capacity, gas concentration, environmental parameters, position information, and failure frequency at time t i ;

[0084] S22, time sequence attention weight calculation: based on the self-attention mechanism, model the time sequence dependence of the input features, and calculate the attention weight matrix A between time points, which is expressed as:

[0085]

[0086] where W q and W k are trainable weight matrices for queries and keys, respectively, and d k is the dimension of the key vector;

[0087] S23, conflict probability prediction output: use the attention weight matrix A as the weight to obtain the value vector sequence V = XWv The weighted sum is performed to obtain the context aggregation representation, and the prediction target time point t n+1 ,t n+2 ,…,t n+m , the communication conflict probability is output using the full connection prediction layer, which is represented as:

[0088] Z=A·(XW v );

[0089] Wherein, W v is a value vector mapping matrix, and Z is the aggregated time sequence context feature;

[0090]

[0091] Wherein, z n is the aggregated context representation of the current time point, w c,k , b c,k are linear parameters for predicting the kth future time slot, σ is a Sigmoid function, and P conflict (t n+k ) is the NB-IoT access conflict probability at the predicted time point t n+k .

[0092] The adaptive power consumption window division in S3 includes:

[0093] S31, power consumption state selection based on conflict probability: receiving the output communication conflict probability sequence, dividing the congestion level of the current period according to the preset congestion classification threshold, including low congestion, medium congestion or high congestion, and according to the congestion level, the NB-IoT module of the intelligent fruit box will switch to the corresponding power consumption working state;

[0094] S32, dynamic asynchronous communication window division: combining the local minimum point in the communication conflict probability sequence, i.e. the predicted low conflict communication empty window period, using the sliding window mechanism to determine the next communication module wake-up time period, each wake-up time period is defined as a dynamic asynchronous communication window [t start ,t end ], the length of which is adjusted to be the shortest according to the current power consumption state and the uplink data volume.

[0095] The power consumption state selection based on conflict probability in S31 includes:

[0096] S311, communication conflict probability receiving: receiving the output future time period communication conflict probability sequence Wherein, m is the number of predicted future time points, P conflict (τ1), P conflict (τ2), …, P conflict (τ mare prediction time points τ1, τ2,..., τ m NB-IoT access conflict probability;

[0097] S312, congestion level division: set the communication congestion probability grading threshold set Θ = {θ L , θ M}, wherein θ L = 0.35 is the lower congestion threshold, θ M = 0.65 is the upper congestion threshold, according to the current conflict probability P conflict (τ cur ) and the congestion grading threshold, the congestion level is divided, specifically including:

[0098] If P conflict (τ cur )≤θ L , it is determined as low congestion;

[0099] If θ L <P conflict (τ cur )≤θ M , it is determined as medium congestion;

[0100] If P conflict (τ cur )>θ M , it is determined as high congestion;

[0101] S313, power consumption state switching execution: according to the current congestion level to control the power consumption state of NB-IoT communication module, specifically including:

[0102] If the current congestion level is low congestion, enter the light wake-up monitoring state (short interval polling + fast access);

[0103] If the current congestion level is medium congestion, maintain the periodic wake-up state (fixed interval polling);

[0104] If the current congestion level is high congestion, switch to deep sleep state (turn off the radio module).

[0105] The dynamic asynchronous communication window division in S32 includes:

[0106] S321, local low conflict time point extraction: based on the output communication conflict probability sequence The neighborhood sliding comparison method is used to extract the time point set satisfying the local minimum condition , denoted as:

[0107]

[0108] Wherein, τ kP(k) is the kth predicted time conflict k conflict k-1 conflict k+1 k k-1 k+1

[0109] S322, low collision section sliding window identification: set a time sliding window radius δ t with each local minimum time point as the center, define the candidate low collision section as If the maximum collision probability in the window satisfies , it is considered as an effective communication empty window, and the window that meets the condition is included in the available communication time period set

[0110] S323, window length minimization adjustment: for each available window According to the current NB-IoT module power consumption state identifier λ mode ∈{low congestion, medium congestion, high congestion} and the total amount of data packets to be uploaded S data , calculate the dynamic minimum communication window length Finally, select the window that meets from the set as the next communication wake-up time period [t wake , t sleep ], and the dynamic minimum communication window length is expressed as:

[0111]

[0112] Where R eff (λ mode ) is the effective upload rate (bytes / second) of the NB-IoT module in different power consumption states, which is obtained according to the module specification or historical data statistics.

[0113] The time slot optimal matching scheduling in S4 includes:

[0114] S41, candidate time slot priority sorting: in the determined available communication time period set , according to the communication collision probability P conflict (τ k ​​​​​​​​​), historical communication success rate, terminal wake-up cost to build a comprehensive priority function, and score all available communication time slots to make the time slots with high predicted success rate and low power consumption rank first.

[0115] S42, resource pool matching and access path construction: obtain the available physical resource block (PRB) set and subcarrier resource pool information broadcast by the current NB-IoT cell, bind and match the high-priority candidate time slots with the idle spectrum in the resource pool, consider resource continuity, terminal access rate and scheduling delay constraints during the matching process, and generate an optimal access plan including the target wake-up time, allocated resource unit and transmission duration using a minimum power consumption path search strategy, to finally form a communication trigger instruction.

[0116] The candidate time slot priority sorting in S41 includes:

[0117] S411, multi-dimensional feature extraction: from the output available communication time period set extract the key features of each candidate window, including the window center point τ k , the communication conflict probability p k = P conflict (τ k ), the historical average success rate γ k ∈ [0, 1] of past uplink communication near the time period (within ±Δ seconds), the energy consumption ε (k) required by the terminal to wake up and complete communication within the window W k , and each window W (k) corresponds to a feature triple (p k , γ k , ε k );

[0118] S412, priority scoring function construction: define a comprehensive priority scoring function Ψ k for each candidate window W (k) , which is expressed as:

[0119]

[0120] where p k is the communication conflict probability (the smaller the better), γ k is the historical success rate (the larger the better), ε k is the wake-up power consumption, ε max = {ε1,..., ε N}, N is the number of available communication windows, and α1, α2, α3 are the weighting coefficients of the scoring function.

[0121] S413, priority sorting and output: according to the scoring value Ψ kright Sort all candidate windows in descending order to form a priority list

[0122] The resource pool matching and access path construction in S42 include:

[0123] S421, obtain cell resource information: the terminal device periodically monitors the broadcast channel (NPRS) of the NB-IoT cell where it is located and obtains the set of physical resource blocks in the current scheduling period and the subcarrier resource pool, expressed as:

[0124]

[0125] in, is a set of physical resource blocks, r1, r2, ..., r P are the 1st, 2nd, ..., Pth available resource blocks respectively, where P is the total number of available resource blocks in the current scheduling period;

[0126] r i =(f i ,t i );

[0127] Among them, r i is the i-th available resource block, f i is the corresponding frequency domain subcarrier number, t i is the time domain slot number;

[0128] S422, high priority time slot matching and path scoring: from the output priority list In the process, try to connect to the physical resource block set in sequence Matching, constructing path scoring function Φ j Represents the power consumption cost of each candidate access path, and selects Φ j The smallest path is the optimal access path Path scoring function φ j Expressed as:

[0129]

[0130] in, is the total transmission time corresponding to the path, The maximum available time allowed for the communication window, is the number of non-continuous frequency domain segments in the path, L total is the total frequency domain length occupied by the path, is the scheduling delay from the current time to the time when the path is available, β1, β2, and β3 are the corresponding weighting coefficients respectively;

[0131] S423, trigger instruction generation: the optimal access path The mapping is a communication scheduling instruction, including a wake-up timestamp, a set of subcarrier numbers used, a set of time slot numbers used, and an expected transmission duration.

[0132] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0133] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A low-power communication method for an intelligent rubbish bin based on NB-IoT, characterized in that: The following steps are involved: S1, asynchronous sensing and collection: This system obtains status data inside the garbage bin, including garbage volume, gas concentration, temperature and humidity. It also combines the device's current geographic location and uplink failure history to create a timestamped asynchronous communication attempt log. S2, time slot conflict prediction modeling: asynchronous communication attempt records are input into the NB-IoT access conflict prediction model based on the temporal attention mechanism to predict the congestion level of future communication time slots and output the communication conflict probability in the target period; S3, Adaptive Power Consumption Window Division: This adjusts the NB-IoT module's power consumption and access window based on the probability of communication conflicts. It uses a non-fixed-cycle asynchronous time-slicing mechanism to activate the communication module during predicted low-congestion time slots, setting the shortest available communication window. S4, optimal time slot matching scheduling: This combines the output of the conflict prediction model with the available subcarrier resource pool provided by the NB-IoT network to perform time slot priority sorting and matching, build an access plan for the minimum power consumption path, and output an optimized communication triggering plan; S5, dynamic execution: According to the communication trigger plan, the NB-IoT communication module is started to send the target data and complete the current round of communication tasks.

2. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 1, characterized in that: The asynchronous sensing acquisition in S1 includes: S11, synchronous acquisition of sensor data: Utilize multiple types of low-power sensors placed inside the garbage bin to obtain status data, including: Garbage capacity data D v :The vertical distance d between the garbage surface and the bin opening is measured by the ultrasonic ranging unit bin ; Gas concentration C g : Target gas concentration is measured by multi-gas sensors; Temperature T s and humidity H s : The current cabinet environment status is obtained by the environmental sensor; S12, device location positioning: call the built-in GNSS module to obtain the current geographical location information of the garbage bin P lat ,P lon And timestamp t pos , forming the position information vector, L pos =(P lat ,P lon ,t pos ); S13, Uplink Failure Event Statistics and Modeling: Traverse the device communication logs, in the time window [t i ,t j ], count the frequency of failure events in NB-IoT communication attempts, and construct the uplink failure frequency function F fail (t k ); S14, build asynchronous communication attempt record: summarize the obtained status data, location information vector and uplink failure frequency, and calculate the result by timestamp t r Building asynchronous communication attempts to log R try (t r ).

3. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 2, characterized in that: The time slot conflict prediction modeling in S2 includes: S21, history encoding: the constructed asynchronous communication attempt record sequence {R try (t1),R try (t2),...,R try (t n )} perform linear encoding to obtain the input feature sequence matrix X=[x1,x2,…,x n ], where each vector x i Including garbage capacity, gas concentration, environmental parameters, location information and failure frequency at time t i The encoding representation of ; S22, temporal attention weight calculation: Based on the self-attention mechanism, the temporal dependency model of the input feature sequence is performed, and the attention weight matrix A between each time point is calculated; S23, conflict probability prediction output: use the attention weight matrix A as the weight pair vector sequence V = XW v Perform weighted summation to obtain the context aggregation representation and predict the target time point t n+1 ,t n+2 ,…,t n+m , a fully connected prediction layer is used to output the communication conflict probability.

4. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 1, characterized in that: The adaptive power consumption window division in S3 includes: S31, power consumption state selection based on conflict probability: Receive the output communication conflict probability sequence, classify the congestion level of the current time period according to the preset congestion classification threshold, including low congestion, medium congestion, or high congestion, and switch the NB-IoT module of the smart fruit peel box to the corresponding power consumption working state according to the congestion level; S32, dynamic asynchronous communication window division: Combined with the local minimum point in the communication conflict probability sequence, that is, the predicted low-conflict communication window period, a sliding window mechanism is used to determine the wake-up time period of the next communication module. Each wake-up time period is defined as a dynamic asynchronous communication window [t start ,t end ], and its length is minimized according to the current power consumption status and the amount of uplink data.

5. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 4, characterized in that: The power consumption state selection based on the conflict probability in S31 includes: S311, communication conflict probability reception: receiving the output communication conflict probability sequence for the future period Where m is the number of predicted future time points, P conflict (τ1),P conflict (τ2),...,P conflict (τ m ) are the predicted time points τ1, τ2, ..., τ m NB-IoT access conflict probability; S312, congestion level classification: set the communication congestion probability classification threshold set Θ = {θ L ,θ M }, where θ L is the lower limit of the congestion threshold, θ M is the upper limit of the congestion threshold, based on the current conflict probability P conflict (τ cur ) is compared with the congestion classification threshold to classify the congestion level, including: If P conflict (τ cur )≤θ L , it is judged as low congestion; If θ L <P conflict (τ cur )≤θ M , it is judged as medium congestion; If P conflict (τ cur )>θ M , it is determined to be highly congested; S313, power consumption state switching execution: Control the power consumption working state of the NB-IoT communication module according to the current congestion level, specifically including: If the current congestion level is low, it enters the light wake-up monitoring state; If the current congestion level is medium, the periodic wake-up state is maintained; If the current congestion level is high, it switches to deep sleep state.

6. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 5, characterized in that: The dynamic asynchronous communication window division in S32 includes: S321, Extraction of local low-conflict time points: Communication conflict probability sequence based on output The neighborhood sliding comparison method is used to extract the time point set that meets the local minimum condition. Expressed as: Among them, τ k is the kth prediction moment, P conflict (τ k ), P conflict (τ k-1 ), P conflict (τ k+1 ) are respectively the time τ k , τ k-1 , τ k+1 The probability of communication conflict; S322, low conflict segment sliding window identification: at each local minimum time point As the center, set the time sliding window radius δ t , define the candidate low-conflict segment as In the window If the maximum conflict probability satisfies It is considered as an effective communication window period, and the window that meets the conditions Included in the set of available communication time slots S323, window length minimization adjustment: for each available window According to the current NB-IoT module power consumption status mark λ mode ∈{low congestion, medium congestion, high congestion} and the total amount of data packets to be uploaded S data , calculate the dynamic minimum communication window length Finally, from the collection Select the earliest one to meet The window is used as the next communication wake-up time period [t wake ,t sleep ].

7. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 6, characterized in that: The optimal time slot matching scheduling in S4 includes: S41, candidate time slot priority sorting: in the determined available communication time period set In the example, according to the communication conflict probability P of each window center point conflict (τ k ), historical communication success rate, and terminal wake-up cost to build a comprehensive priority function, score and sort all available communication time slots, so that the time slots with high predicted success rate and low power consumption are ranked higher; S42, resource pool matching and access path construction: Obtain the available physical resource block set and subcarrier resource pool information broadcast by the current NB-IoT cell, bind and match high-priority candidate time slots with idle spectrum in the resource pool, consider resource continuity, terminal access rate and scheduling delay constraints during the matching process, and use the minimum power consumption path search strategy to generate the optimal access plan, including the target wake-up time, allocated resource units and transmission duration, and finally form a communication trigger instruction.

8. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 7, characterized in that: The candidate time slot priority sorting in S41 includes: S411, multi-dimensional feature extraction: from the output of the available communication time period set In the above example, the key features of each candidate window are extracted, including the window center point τ k The communication conflict probability p k =P conflict (τ k ), the historical average success rate of uplink communication in the past around this time period γ k 、Terminal in window W (k) Energy consumption required to wake up and complete communication k , each window W (k) Corresponding to a feature triple (p k ,γ k ,ε k ); S412, Priority scoring function construction: define comprehensive priority scoring function Ψ k For each candidate window W (k) Scoring is expressed as: Among them, ε max ={ε1,...,ε N }, N is the number of available communication windows, α1, α2, α3 are the weighting coefficients of the scoring function; S413, priority sorting and output: According to the score value Ψ of each window k right Sort all candidate windows in descending order to form a priority list 9. The low-power communication method for an intelligent rubbish bin based on NB-IoT according to claim 8, characterized in that: The resource pool matching and access path construction in S42 include: S421, obtain cell resource information: the terminal device periodically monitors the broadcast channel of the NB-IoT cell where it is located and obtains the set of physical resource blocks in the current scheduling period and subcarrier resource pool; S422, high priority time slot matching and path scoring: from the output priority list In the process, try to connect to the physical resource block set in sequence Matching, constructing path scoring function Φ j Represents the power consumption cost of each candidate access path, and selects Φ j The smallest path is the optimal access path S423, trigger instruction generation: the optimal access path The mapping is a communication scheduling instruction, including a wake-up timestamp, a set of subcarrier numbers used, a set of time slot numbers used, and an expected transmission duration.

Images