Intelligent electric meter dual-mode communication adaptive switching method

Abstract

The application relates to the technical field of digital communication, in particular to a smart meter dual-mode communication adaptive switching method, the application proposes the following scheme: dividing running data into multiple data units and performing priority evaluation at a sending end, positioning a clearing cut point in a priority sequence in combination with the time delay, packet loss, energy consumption and bandwidth state of two links, so that differentiated allocation of data between a high-speed link and a low-speed link is realized; when data is allocated to the low-speed link, sparse coding is performed by using a preset dictionary, and differential update is triggered when an error exceeds a limit, so that decoding consistency is ensured; the receiving end realizes unified analysis of original messages and sparse messages through cross-link association, dictionary update and residual error synthesis; the application can guarantee the timeliness of key messages, reduce communication energy consumption and cost, and improve the collaborative efficiency of dual-mode links.

Description

Technical Field

[0001] This application relates to the field of digital communication technology, and in particular to a method for adaptive switching of dual-mode communication in smart meters. Background Technology

[0002] Current smart meter data acquisition primarily relies on a single communication link, or employs a session-level, full-segment handover strategy in dual-mode communication scenarios. In this mode, critical messages may be delayed or lost due to untimely handover, while large volumes of ordinary data may occupy high-speed link resources for extended periods, resulting in uneven resource allocation. On the other hand, while low-speed links offer advantages in stability and free data transmission, they are often in standby mode in practical applications, with low utilization rates, making it difficult to handle large-scale data transmission demands. Furthermore, when using low-speed links for data transmission, the receiving end often lacks accurate message parsing, easily leading to reconstruction errors or data inconsistencies. Overall, existing technologies struggle to simultaneously address the timeliness of critical data, the economic efficiency of ordinary data transmission, and cross-link data consistency, thus failing to effectively realize the potential of dual-mode communication.

[0003] To address the above issues, this application proposes an adaptive switching method for dual-mode communication in smart meters. Summary of the Invention

[0004] The technical problem this application aims to solve is to address the shortcomings of existing technologies by providing a dual-mode communication adaptive switching method for smart meters. At the transmitting end, the operating data is divided into multiple data units and prioritized. Combining the latency, packet loss, energy consumption, and bandwidth status of the two links, a clearing cut point in the priority sequence is located to achieve differentiated data allocation between the high-speed and low-speed links. When data is allocated to the low-speed link, sparse encoding is performed using a preset dictionary, and differential updates are triggered when errors exceed limits to ensure decoding consistency. At the receiving end, unified parsing of the original message and the sparse message is achieved through cross-link association, dictionary updates, and residual synthesis.

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] A dual-mode communication adaptive switching method for smart meters is applied to the transmitting end of a communication system. The communication system includes a first communication link and a second communication link, where the transmission rate of the second communication link is lower than that of the first communication link. The communication system also includes a receiving end, which is communicatively connected to the transmitting end via the first and second communication links. The transmitting end and the receiving end are configured with the same preset dictionary set. The method includes:

[0007] The data to be sent is divided into multiple data units, each of which includes at least one message.

[0008] Obtain the link status parameters of the first and second communication links, and determine the target communication link corresponding to each data unit based on the priority of the data unit. Determining the target communication link includes:

[0009] The priorities of multiple data units are aggregated to obtain a priority sequence. Based on the link status parameters, the clearing cut point in the priority sequence is located. Data units in the priority sequence that are before the clearing cut point are assigned to the first communication link, and data units in the priority sequence that are at or after the clearing cut point are assigned to the second communication link.

[0010] If the target communication link corresponding to the data unit is the second communication link, the encoding information of the data unit is generated through a sparse strategy, and the encoding information is forwarded through the second communication link.

[0011] If the target communication link corresponding to the data unit is the first communication link, the corresponding data unit is sent to the receiving end through the first communication link.

[0012] The link status parameters include at least one of latency, packet loss rate, bandwidth utilization, and energy consumption indicators.

[0013] The priority is calculated in the following ways:

[0014] The messages in the data unit are characterized to generate a summary layer for representing business semantics and multiple residual layers with increasing information granularity.

[0015] The summary layer is prioritized according to a preset business semantic strategy library to obtain the initial priority of the data unit;

[0016] The effect index of each residual layer is calculated, and the effect index is synthesized into an effect score through a preset mapping function. The effect index includes at least one of the following: the marginal information gain of the residual, the substitutability index with historical samples, the remaining time margin, and the estimated transmission energy consumption.

[0017] The priority of a data unit is obtained based on the initial priority and the effect score, combined with a preset grading threshold.

[0018] Determining the target communication link corresponding to each data unit includes:

[0019] The priorities of multiple data units are aggregated to obtain a priority sequence, wherein the priority sequence is sorted in descending order according to the priority of the data units;

[0020] Based on the priority sequence and in conjunction with the link state parameters of the first and second communication links, the clearing cut point in the priority sequence is located. Data units in the priority sequence located before the clearing cut point are allocated to the first communication link, while data units in the priority sequence located after the clearing cut point and data units corresponding to the clearing cut point are allocated to the second communication link.

[0021] The liquidation cut points in the location priority sequence include:

[0022] Within a preset prediction time window, the shadow prices corresponding to the first communication link and the second communication link are calculated based on the link status parameters, and the available service budget of the first communication link in the prediction time window is determined.

[0023] For each data unit in the priority sequence, calculate the marginal utility density of the data unit under unit transmission overhead;

[0024] Using the difference between the shadow prices corresponding to the first and second communication links as the liquidation threshold, the transmission overhead of data units is accumulated sequentially from the beginning of the priority sequence until the accumulated value is less than the available service budget, and the marginal utility density of the current data unit is greater than or equal to the liquidation threshold, at which point the current data unit is designated as the liquidation cut point.

[0025] The step of generating the encoding information of the data unit using a sparse strategy includes:

[0026] The data unit is type-identified, and a target dictionary is selected from the preset dictionary set according to the identification result, wherein the target dictionary includes version identifier and fingerprint verification information;

[0027] The data unit is sparsely encoded and reconstructed locally according to the target dictionary to obtain a sparse location set and quantization coefficients, and the error estimate is calculated based on the local reconstruction result.

[0028] If the error estimate is greater than the preset error threshold, the preset dictionary set is updated according to the error estimate to obtain a dictionary difference update package;

[0029] Generate an encoding header corresponding to the data unit based on the dictionary differential update packet, the sparse location set, and the quantization coefficients, and send a transmission message including the encoding header to the receiving end through the second communication link.

[0030] A dual-mode communication adaptive switching method for smart meters is applied to the receiving end of a communication system. The communication system includes a first communication link and a second communication link, where the transmission rate of the second communication link is lower than that of the first communication link. The communication system also includes a transmitting end, which is communicatively connected to the receiving end via the first and second communication links. The transmitting end and the receiving end are configured with the same preset dictionary set. The method includes:

[0031] Receive a first type of transmission message from the first communication link and a second type of transmission message from the second communication link, wherein the first type of transmission message is an original message and the second type of transmission message is a sparsely coded message;

[0032] For the second type of transmission message, determine whether it carries indication information of dictionary differential update packet;

[0033] If so, the preset dictionary set is updated according to the dictionary difference update package to obtain the updated dictionary set;

[0034] The second type of transmission message is parsed according to the updated dictionary set to obtain the corresponding data unit reconstructed message;

[0035] The data unit reconstructed message and the first type of transmission message are combined to obtain the final message.

[0036] The method further includes establishing a correspondence between the second type of transmission messages and the first type of transmission messages, specifically including:

[0037] At the receiving end, a cross-link association ledger is constructed, wherein the cross-link association ledger is partitioned and managed according to the time grid of the first type of transmission message;

[0038] For the first type of transmission message, an alignment binding code for cross-link alignment is extracted, and the first type of transmission message is stored in the cross-link association ledger according to the alignment binding code, wherein the alignment binding code is obtained by combining the transaction sequence number, time index and semantic anchor set calculated based on the original message of the first type of transmission message;

[0039] For the second type of transmission message, an alignment candidate binding code for cross-link alignment is extracted, wherein the alignment candidate binding code is obtained by combining the transaction sequence number, time index, and sparse support digest of the second type of transmission message;

[0040] Based on the delay distribution of the second communication link, a time dilation window matching the alignment candidate binding code is determined, and the cross-link association ledger is retrieved based on the time dilation window to obtain multiple candidate grids, wherein the candidate grids include at least one type of first-class transmission message.

[0041] Based on the alignment candidate binding code and the corresponding alignment binding code in the candidate grid, the first type of transmission message corresponding to the second type of transmission message is determined.

[0042] The step of updating the preset dictionary set according to the dictionary difference update package includes:

[0043] The dictionary differential update package is subjected to integrity and source authentication, and the target version identifier, parent version fingerprint, segment index and atomic operation sequence are parsed.

[0044] The consistency of the preset dictionary set is verified based on the parent version fingerprint. If there is a discrepancy, the preset dictionary set is updated. The update includes: adding, deleting, replacing, rearranging and mapping, adjusting weights, merging subspaces and splitting subspaces of the preset dictionary set according to the atomic operation sequence and segment index, and gradually generating a verification hash. The version identifier of the preset dictionary set is modified according to the target version identifier.

[0045] Compared with the prior art, the beneficial effects of this application are:

[0046] This application constructs a priority-sequence-based clearing cut-point mechanism to dynamically allocate data of different values ​​between high-speed and low-speed links. This ensures that critical messages receive priority access to high-speed links for timely delivery, while ordinary, large-volume data is efficiently transmitted in low-speed links using sparse coding, thus balancing transmission efficiency and cost control. By introducing dictionary differential updates and cross-link association mechanisms, the receiving end can accurately reconstruct messages without relying on full transmission, ensuring data consistency and integrity. This application reduces the number of cellular link wake-ups and energy consumption while improving the utilization rate of the power grid, ultimately achieving stability, economy, and long-term sustainability of smart meter data collection in a dual-mode communication environment. Attached Figure Description

[0047] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0048] Figure 1 This is a schematic diagram illustrating an exemplary application scenario of an embodiment of this application;

[0049] Figure 2 This is a schematic diagram illustrating the principle of a solution according to an embodiment of this application;

[0050] Figure 3 This is a schematic diagram of the communication system modules according to an embodiment of this application;

[0051] Figure 4This is a flowchart illustrating an adaptive switching method for dual-mode communication in a smart meter according to an embodiment of this application.

[0052] Figure 5 This is a schematic diagram of the priority sequence of embodiments of this application;

[0053] Figure 6 This is a schematic diagram illustrating the principle of clearing cut point positioning in an embodiment of this application;

[0054] Figure 7 This is a flowchart illustrating another method for adaptive switching of dual-mode communication in a smart meter according to an embodiment of this application. Detailed Implementation

[0055] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0056] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0057] This application is proposed for dual-mode communication scenarios in intelligent electricity information collection systems and is applicable to field environments where wireless communication and dedicated power networks coexist and the link capabilities and cost models are significantly asymmetrical.

[0058] It is understood that the wireless communication refers to cellular communication networks built by public mobile communication operators, such as public cellular networks, Wi-Fi, 2.4G, Bluetooth, Zigbee, LoRa, and sub-1G. These networks are characterized by wide coverage and relatively high bandwidth, but they will incur data charges and wake-up overhead in the long run.

[0059] The so-called power private network refers to a communication network built by the power system itself or dedicated to the industry, such as a power line carrier communication network, a distribution automation fiber optic network, or a power industry wireless private network. Its characteristics are that no additional traffic costs are required and it has a relatively stable access relationship. However, it is limited by bandwidth, latency, and anti-interference capabilities, and its transmission rate is usually lower than that of wireless communication.

[0060] In this application, wireless communication is the first communication link, and the dedicated power grid is the second communication link.

[0061] The consensus in the industry is that the amount of data in routine meter reading is not huge; however, in actual operation and maintenance, the amount of data is not the only bottleneck. When to use which chain and how to deliver data stably under the constraints of energy consumption and cost are more common and more difficult engineering dilemmas.

[0062] For example, typical time periods include centralized reporting of fee control settlement windows, alternation of demand assessment slices, grid connection disturbances of distributed power sources, or alarm storms in the distribution area. At this time, the cellular side may jitter due to cell congestion, while the private network side is often affected by carrier noise, inbound wiring and metal meter boxes, resulting in sudden increases in latency, uneven packet loss, or excessive wake-up overhead.

[0063] It is easy to understand that existing communication methods are often forced to oscillate between a one-size-fits-all approach at the session level and frequent handshakes that lead to a sharp increase in handshake costs. This affects real-time control and cost control receipts, and also reduces the lifespan of batteries and supercapacitor meters. When the scale expands to tens of thousands of terminals, even if the data of a single meter is not large, the systemic costs of switching and retransmission will be amplified and become a continuous burden on the master station and the access side.

[0064] Given the aforementioned premises, this application starts with the link selection at the sending end and provides a more robust path selection logic. The core principle is as follows:

[0065] The information to be transmitted is extracted into message information that can directly drive settlement and operation and maintenance decisions according to business semantics, and details that have a minor impact on the results are folded into normal information. On this basis, the sending end and the receiving end maintain a lightweight consensus dictionary, so that normal information is transmitted with sparse coefficients on a low-speed, low-power second communication link, while occasional dictionary differences and messages that need to be transmitted quickly are only carried in small quantities in the available window of the first link.

[0066] Understandably, this application does not pursue the ultimate compression rate, but rather transforms the implicit costs of frequent switching and repeated handshakes into low-frequency consistency maintenance, thereby achieving latency stability and energy consumption certainty across the entire network.

[0067] This application applies to household electricity meters located in basements and metal distribution boxes in urban areas, centralized meters in mountainous areas with long wiring, intermittent online terminals powered by batteries, and groups of distribution stations during periods of grid fluctuation or high alarm frequency.

[0068] The common characteristics of the applicable objects are asymmetric link capabilities, limited collection perspective, occasional superposition of alarms or receipts, and hard constraints on data consistency imposed by compliance audits.

[0069] The sending strategy proposed in this application rearranges the timing of some data to be transmitted via the fast chain and some via the slow chain without changing the business process on the main station side. This avoids the oscillation and energy waste of session-level switching and fits the real rhythm of long-term operation and maintenance on the power distribution side.

[0070] refer to Figure 1 , Figure 1 This is a schematic diagram illustrating an exemplary application scenario provided in an embodiment of this application.

[0071] Figure 1 The diagram shows multiple smart meters distributed at electricity consumption sites, which need to transmit meter reading data and event information to the data center on the access side via a communication network.

[0072] Figure 1 The data shows that all messages converge along a single communication path. The meter does not differentiate between message types, information values, or network load status. This approach may be sufficient for ordinary periodic meter reading scenarios, but when alarm reports, fee control instruction receipts, and large-volume curve data transmissions occur simultaneously, the bandwidth and access stability of the cellular link are difficult to match the diverse needs of various services.

[0073] On the other hand, although the dedicated power grid is Figure 1 While backup paths exist, current technologies often designate them as remedial links after failures, leaving them almost idle during normal business operations, resulting in low resource utilization. This creates an asymmetry between link utilization and business needs: critical instructions may be delayed and transmitted after low-priority data, while large volumes of low-value data occupy high-cost, high-power cellular links for extended periods.

[0074] refer to Figure 2 , Figure 2 This is a schematic diagram illustrating the principle of a solution provided in an embodiment of this application.

[0075] It is understandable that this application does not simply switch wireless communication and power grid as primary / backup links, but rather differentiates data units.

[0076] exist Figure 2 In embodiments not shown, the smart meter first parses the data to be transmitted, dividing it into key data units with high timeliness or strong consistency requirements, and ordinary data units that are larger in volume but have lower value density.

[0077] In this way, wireless communication is no longer occupied by low-value data for a long time, and the power grid is no longer in a simple standby state, but forms a dynamic complement to wireless communication.

[0078] refer to Figure 3 , Figure 3 A schematic diagram of the communication system provided in an embodiment of this application.

[0079] Figure 3 The communication system shown includes a transmitter and a receiver, both configured with the same preset dictionary set, wherein:

[0080] The sending end includes a data unit partitioning module, a link allocation module, and a data compression and transmission module.

[0081] The data unit partitioning module is used to split the data to be transmitted into multiple data units and perform feature processing.

[0082] The link allocation module is used to determine the target communication link based on the link status parameters and priority sequence.

[0083] The data compression and transmission module is used to perform sparse coding based on the target dictionary to generate a sparse location set and quantization coefficients when the data unit is allocated to the power grid, and to send the encoded information through the power grid. When the data unit is allocated to wireless communication, the original message is sent directly.

[0084] The receiving end includes a data receiving module, a data matching module, and a data restoration module.

[0085] The data receiving module is used to receive first-class and second-class messages transmitted via different links.

[0086] The data matching module is used to establish a mapping relationship in the cross-link association ledger based on the dictionary version identifier, transaction sequence number and time index in the message header, and to trigger dictionary differential update when necessary.

[0087] The data restoration module decodes and reconstructs the sparse parameters based on the updated dictionary set, and then synthesizes them with the first type of message to obtain the final message.

[0088] It is easy to understand that, Figure 3 The modular structure shown embodies the core idea of ​​this application: through differentiated link allocation and sparse compression at the sending end, version consistency maintenance and cross-link message synthesis at the receiving end, efficient and stable data transmission in the dual-mode communication scenario of smart meters can be achieved.

[0089] Next, combined Figure 4 This application introduces a dual-mode communication adaptive switching method for smart meters provided in its embodiments. Figure 4 The method shown is applied to the transmitting end of a communication system, which includes a first communication link and a second communication link, wherein the transmission rate of the second communication link is less than that of the first communication link. The communication system also includes a receiving end, which is communicatively connected to the transmitting end through the first and second communication links. The transmitting end and the receiving end are configured with the same preset dictionary set. The specific steps of the method are as follows:

[0090] S1: Divide the data to be sent into multiple data units, each data unit including at least one message;

[0091] In this embodiment, the division of data units can be based on multi-source data such as periodic meter reading data, event alarm information, fee control instruction receipts, and operation logs.

[0092] The types of multi-source data can be flexibly configured according to actual applications, as long as they can support the determination of the meter's operating status to a minimum. This application does not impose too many restrictions on this.

[0093] This approach avoids the one-size-fits-all processing of entire batches of messages and allows subsequent link allocation to be adjusted at the data unit level, thus enabling more flexible use of dual-link resources in diverse business scenarios.

[0094] S2: Obtain the link status parameters of the first and second communication links, and determine the target communication link corresponding to each data unit by combining the priority of the data unit.

[0095] In this embodiment, the link status parameters may include any one or more of latency, packet loss rate, bandwidth utilization, or energy consumption indicators. The specific collection method can be obtained through existing link detection methods, which will not be elaborated here.

[0096] Those skilled in the art will understand that by combining link parameters with data unit priorities to form a priority sequence, and further locating the clearing cut point, data units before the cut point are allocated to the first communication link, and data units after the cut point are allocated to the second communication link. This ensures the timeliness of critical message transmission while reducing the risk of redundant transmission on low-speed links, and avoids the delay and energy waste caused by coarse-grained switching in traditional methods.

[0097] S3: If the target communication link corresponding to the data unit is the second communication link, the encoding information of the data unit is generated through a sparse strategy, and the encoding information is forwarded through the second communication link;

[0098] In this embodiment, the encoded information is obtained by performing sparse coding on a target dictionary selected from a preset dictionary set, resulting in a sparse location set and quantization coefficients. Dictionary differential update packets are generated when necessary to maintain consistency between the receiver's and sender's dictionaries. This allows low-speed links to transmit lightweight data in the form of sparse parameters instead of the complete original message, significantly reducing transmission latency and energy consumption. Especially in scenarios where bandwidth is limited in power grid private networks, effective data unit restoration can still be guaranteed, solving the problem that in existing solutions, low-speed links can only serve as backup channels.

[0099] S4: If the target communication link corresponding to the data unit is the first communication link, the corresponding data unit is sent to the receiving end through the first communication link;

[0100] In this embodiment, when the link determination result is the first communication link, the data sent can be the original message. When critical messages must arrive within a specified time limit, directly transmitting the original message can avoid errors and delays introduced by compression and restoration. This hierarchical processing method effectively balances the resource contradiction between high-time-sensitive services and low-value services, realizing differentiated scheduling in dual-mode communication scenarios.

[0101] Before detailing the specific technical aspects of the steps, this application's embodiments need to reiterate:

[0102] While smart meters typically handle small amounts of data per transaction in daily applications, link selection and energy balance are crucial factors affecting system stability in scenarios involving concurrent services or frequent network fluctuations. Traditional methods often operate on a session-by-session basis, leading to inevitable delays and retransmissions if the path is incorrect. In contrast, this application's implementation logic does not treat communication links as a simple primary / backup relationship. Instead, it constructs a priority sequence and cutpoint logic to link link status with data utility, enabling the two links to complement each other within the same time window.

[0103] In its implementation, this application abstracts the utility of a data unit as the ratio of information gain to energy consumption cost, and uses this ratio to form a decreasing distribution curve within a time window. Changes in link status do not directly trigger an overall switchover; instead, stable intersection points are found on the curve to determine the boundaries, ensuring that critical data in the first half is carried by high-speed links, while the remaining portion is transmitted on low-speed links in the form of sparse coefficients. The focus of the processing logic is not on maximizing the compression ratio, but on eliminating the oscillations caused by frequent switching, allowing link usage to naturally stratify according to business value.

[0104] Furthermore, this application also considers dictionary update behavior in conjunction with link scheduling. When the reconstruction error of sparse decoding exceeds the threshold, the sender does not immediately abandon the low-speed link, but inserts the dictionary differential update packet as a special data unit into the transmission sequence and clears it together with other data.

[0105] Next, we will further elaborate on the technical aspects regarding priority in the method of this application.

[0106] The priority is calculated in the following ways:

[0107] The messages in the data unit are characterized to generate a summary layer for representing business semantics and multiple residual layers with increasing information granularity.

[0108] Specifically, the message undergoes semantic restructuring before entering the priority evaluation stage:

[0109] Extract statistical data from meter reading messages, including electricity consumption during the time period, quantiles, peak and valley extreme values, and limit violation markers;

[0110] Extract the trigger code, duration, and level for events or alarms;

[0111] Extract execution status and failure reason codes from expense control receipts; and perform detrending, period stripping, and change point detection on curve-type data to form summary elements that can directly support settlement and operation judgment.

[0112] Furthermore, the original sequence is mapped to baseline reconstruction, and the residuals are then segmented from low to high information granularity, including: shape correction residuals (low-frequency structural differences), local anomaly residuals (short-term pulses / steps), and detail texture residuals (high-frequency perturbations), to obtain the summary layer and multi-level residual layers.

[0113] In this embodiment, to ensure that different data sources can be processed uniformly, the summary layer adopts a structured key-value layout and fixed-order field packaging, and the residual layer is described by block slices and atomic residual units. The atomic residual unit records the dictionary index range, start and end timestamps and verification fingerprints.

[0114] The summary layer is prioritized according to a preset business semantic strategy library to obtain the initial priority of the data unit;

[0115] Specifically, the business semantic policy library maintains the mapping from different message semantics to basic priorities in the form of rule sets, covering dimensions such as settlement sensitivity, control timeliness, operation and maintenance evidence collection, and compliance retention. For example, fee control execution and abnormal trip receipts belong to the time-controlled category, settlement correction and theft alarms belong to the evidence collection dependency category, and regular curves and operation logs belong to the statistical category.

[0116] Understandably, the construction process of the business semantic strategy library includes the following three aspects:

[0117] In the first aspect, based on the settlement criteria, scheduling / fee control SOP, alarm classification specifications and compliance retention requirements, elements are extracted from historical messages and on-site work orders to form a reusable semantic tagging system and scenario metadata.

[0118] Secondly, atomic rule units are established for each type of semantic tag. Each atomic rule unit includes at least triggering conditions, context conditions, output items, conflict resolution, and inheritance relationships. According to the business chain, atomic rules are arranged into a combination strategy according to the transaction chain. Progressive weights and cross-message associations are set for the preceding and following links of the same transaction, so that the subsequent links are boosted after the preceding links are triggered. At the same time, attenuation curves are configured for repeated statistical messages to avoid occupying the emergency channel.

[0119] Thirdly, the combined strategies are validated and A / B compared using offline regression methods to verify the arrival rate and cellular / private network load sharing ratio during key periods. Once the threshold is met, the strategies are solidified in the form of a strategy package. The strategy package carries a version number, fingerprint, and scope of effectiveness, and can be distributed in a differentiated manner according to the area, device type, or billing period.

[0120] Those skilled in the art will understand that the specific field names, threshold settings, and verification processes can be adjusted according to operational procedures and regional differences, as long as a stable mapping from message semantics to basic priority can be achieved to the minimum extent possible. This application does not impose any further limitations.

[0121] The effect index of each residual layer is calculated, and the effect index is synthesized into an effect score through a preset mapping function. The effect index includes at least one of the following: the marginal information gain of the residual, the substitutability index with historical samples, the remaining time margin, and the estimated transmission energy consumption.

[0122] Specifically, for each residual layer, the marginal contribution to the business objective and the transmission cost incurred are calculated.

[0123] Marginal information gain is assessed by comparing the difference between making a judgment based solely on the summary layer and making a judgment after adding the residual layer. For example, it assesses whether the settlement error has been reduced to an acceptable range, whether the fault location has been narrowed down to a single device, and whether false positives have been eliminated in abnormal reconstruction. The substitutability index is based on local short-term predictions and references to neighboring devices to determine whether the information at this layer can be inferred from historical patterns or neighborhood correlations. The more difficult it is to be inferred as substitutable, the higher the index. The remaining time limit margin is given by the scenario strategy, such as the last available time for compliance retention, ledger reconciliation, and time-period settlement. The estimated transmission energy consumption is assessed based on the message byte size, whether the first link needs to be woken up, and the possible retransmission and handshake overhead.

[0124] In this embodiment, to facilitate stable operation at the edge, the mapping function uses a piecewise decision table instead of a complex model: a high-effect level is directly assigned when the marginal gain is high and substitutability is low; the level is increased when the marginal gain is in the medium range but the remaining timeframe is approaching; the level is decreased when energy consumption estimates exceed the budget and substitutability is high; when the receiver reports a dictionary consistency gap or reconstruction verification fails, the corresponding residual layer is marked as necessary for correction, and its effect level is temporarily increased. This mapping is traceable and verifiable with explicit rules, facilitating the explanation of the judgment basis in procedure audits or acceptance scenarios.

[0125] The priority of the data unit is obtained based on the initial priority and the effect score, combined with the preset grading threshold.

[0126] Specifically, the synthesis process uses data units as carriers and merges the initial priority with the residual layer effect scores: if a data unit contains only a summary layer or its residual layer effects are all at a low level, the initial priority remains unchanged; if any residual layer is required for correction or there is a time urgency, a link tendency flag is set on the data unit, indicating that subsequent scheduling should prioritize high-speed links; if multiple residual layer effects are scattered at different levels, a dominant layer is set according to the decisiveness of the result and the pressure on energy consumption, and the tendency and level of the dominant layer are used to correct the final level of the data unit.

[0127] Next, we will further elaborate on the technical content of the target communication link in the method of this application.

[0128] Understandably, after obtaining the priorities of all data units, it is necessary to allocate data units of different priorities within a limited time window, taking into account the link's operating status, to determine their target communication links. The core of this application is not a simple binary judgment of high priority using fast links and low priority using slow links, but rather a comprehensive modeling approach that integrates link capacity, latency fluctuations, energy consumption budgets, and business closed-loop requirements into a unified computing framework, forming a dynamically updatable allocation logic.

[0129] In one example, determining the target communication link corresponding to each data unit includes:

[0130] S2.1: Summarize the priorities of multiple data units to obtain a priority sequence, wherein the priority sequence is sorted in descending order according to the priority of the data units;

[0131] refer to Figure 5 , Figure 5 This is a schematic diagram of the priority sequence in an embodiment of this application.

[0132] Figure 5 The data to be transmitted consists of five data units, namely data unit one, data unit two, data unit three, data unit four, and data unit five.

[0133] like Figure 5 As shown, the data units are ordered in descending order as follows: Data Unit 4, Data Unit 2, Data Unit 1, Data Unit 5, and Data Unit 3.

[0134] S2.2: Based on the priority sequence and in conjunction with the link status parameters of the first and second communication links, locate the clearing cut point in the priority sequence, allocate the data units in the priority sequence that are before the clearing cut point to the first communication link, and allocate the data units in the priority sequence that are after the clearing cut point and the data units corresponding to the clearing cut point to the second communication link.

[0135] Specifically, in dual-mode communication scenarios, there are inherent differences in the capabilities and costs of wireless communication and dedicated power grids: wireless communication has high bandwidth and low latency, but comes with traffic costs and energy consumption from frequent wake-ups; dedicated power grids are free and have stable access relationships, but have low transmission rates and are prone to backlog when there are large concurrency.

[0136] In this embodiment, a balance point is found as the clearing cutoff point by combining the service budget and overall cost of the two links within the current time window. The clearing cutoff point is both the intersection of service value and link cost, and the watershed between high-priority and low-priority data on the two links, thereby achieving fine-grained scheduling. Data units located before the clearing cutoff point are all allocated to wireless communication to ensure the timeliness and reliability of high-value messages; data units after the clearing cutoff point are carried by the dedicated power grid to leverage its stable and cost-free characteristics; and data units at the clearing cutoff point are allocated to the dedicated power grid by default.

[0137] refer to Figure 6 , Figure 6 This is a schematic diagram illustrating the principle of clearing cut point positioning in an embodiment of this application.

[0138] Figure 6 by Figure 5 Taking the priority sequence as an example, first calculate the liquidation threshold and available service budget. Based on the first data unit four in the priority sequence, determine whether it is less than the liquidation threshold and available service budget. If so, add data unit two. After repeated calculation, confirm that when data unit one is added, it is less than the available service budget but greater than the liquidation threshold. Then, locate data unit one as the liquidation cut point.

[0139] exist Figure 6 In an embodiment not shown, data unit four and data unit two are packaged and assigned to the first communication link, and data unit one, data unit five and data unit three are packaged and assigned to the second communication link.

[0140] Understandably, the judgment criteria in this application may fall into two categories under different scenarios: one is that the cumulative cost is less than the service budget, but the marginal utility density is greater than or equal to the liquidation threshold. In this case, it means that putting the data unit into the first communication link will lead to a decrease in the link's cost-effectiveness, so the data unit is directly used as the liquidation cut point. The other is that the cumulative cost has exceeded the service budget, but the marginal utility density of the data unit is still less than the liquidation threshold. In this case, it means that although the data unit has value, it exceeds the budget's carrying capacity, and the unit is also used as the liquidation cut point. Regardless of the scenario, the essential significance of the cut point is that it is the final intersection point between the two curves of the service cost that the link can carry and the business utility reflected by the data.

[0141] In one example, the liquidation cut point in the location priority sequence includes:

[0142] S2.2.1: Within a preset prediction time window, calculate the shadow prices corresponding to the first communication link and the second communication link based on the link status parameters, and determine the available service budget of the first communication link within the prediction time window;

[0143] Specifically, shadow pricing is used to unify the link costs from different sources and with different dimensions to a comparable benchmark of comprehensive overhead per unit bit, making it easier to align with the service value of a data unit. Within the prediction time window, shadow pricing consists of factors such as transmission latency uncertainty, retransmission risk due to packet loss, handshake overhead required for hold and wake-up, energy consumption per unit bit, and tariff or quota penalties.

[0144] It is important to note that handshake and wake-up are not constant values; they are affected by the activity window, intermittent reception status, and concurrent occupancy, and must be amortized according to the expected burst patterns within the time window. The available service budget characterizes the available capacity of the communication link that can be flexibly scheduled within the predicted time window. It is neither the peak rate nor the instantaneous upper limit of physical layer throughput, but rather an allocable interval after deducting rigid occupancy, retaining necessary redundancy, and considering the activity window structure.

[0145] It is easy to understand that focusing solely on rate or latency will obscure the costs of handshaking and energy consumption, while focusing solely on cost or energy consumption will ignore the business benefits brought by low latency. Only by using both together can an executable scheduling constraint be formed at the sending end.

[0146] In this embodiment, the calculation of the shadow price relies on the link state parameters calculated above:

[0147] By combining passive statistics with lightweight probe messages, the system collects the quantile trajectory of round-trip delay, the sliding statistics of packet loss and retransmission, the time proportion of duty cycle and available rate, the start and end distribution of active time slots and active windows.

[0148] Energy consumption is estimated using a decomposition model based on a static baseline, and one-time costs are amortized back to the unit bit according to the expected burst size; the impact of tariffs or quotas is injected into the shadow price in a tiered penalty manner, so that the cost per bit automatically increases within the time window approaching the threshold.

[0149] Furthermore, the determination of the available service budget follows the principle of hard priority and soft guarantee: first, the rigid occupancy of upstream instructions and guaranteed receipts is deducted from the theoretical carrying capacity, then the minimum heartbeat and keep-alive load required for RRC maintenance is deducted, and a certain proportion of disturbance redundancy is reserved in the remaining part to obtain the flexible budget that can be used for the allocation logic of this application, where RRC stands for Radio Resource Control, specifically a resource control protocol.

[0150] In some special cases, if concurrent services are detected to trigger an active window, the capacity within that window is listed separately as parasitic availability, and only boundary items are allowed to be injected in differential fragments without extending the active window.

[0151] S2.2.2: For each data unit in the priority sequence, calculate the marginal utility density of the data unit under unit transmission overhead;

[0152] Specifically, marginal utility density is used to measure "how much additional benefit can be brought to the business objective for each unit of link resource consumed", establishing a direct correspondence between business semantics and link cost.

[0153] In some optional implementations, the measurement of business benefits is not limited to a single dimension, but is evaluated around four aspects: settlement accuracy, control timeliness, operation and maintenance location capabilities, and compliant retention reliability. If the expected results can be achieved solely by the summary layer, the marginal benefit of that data unit is low. If adding a residual layer can reduce settlement deviations to an acceptable range, narrow the alarm location range from the district level to the device level, or complete the evidence chain closure before the retention period expires, then the marginal benefit of that data unit is considered decisive. Meanwhile, unit transmission overhead includes, but is not limited to, the number of bytes, and also considers the size after sparse coding under the second communication link, the residual compensation size under the first communication link, whether additional wake-up is required, and the number of retransmissions and handshakes that may be triggered.

[0154] In this embodiment, the marginal utility density of each data unit is obtained through a rule table, specifically including the following two dimensions:

[0155] In the first dimension, the role of the data unit in the current transaction stage is determined based on the business semantic strategy library. Then, combined with the effect index of the residual layer, it is determined to belong to one of the three categories: decisive, supporting, or modifying. Decisive items are directly assigned high density, supporting items are set with intermediate density based on the remaining time limit and substitutability index, and modifying items are adjusted within the low density range.

[0156] In the second dimension, when the estimated transmission overhead reaches the energy consumption limit or the activity window may be extended, the density under the same business revenue is appropriately reduced according to energy consumption sensitivity. To avoid short-term fluctuations causing density levels to jump back and forth, a dual-threshold control with different entry / exit thresholds is adopted, and boundary labels are preset for data units near the threshold, so that they can be allocated using mild methods such as fragmentation parallelism and parasitic injection in subsequent allocation.

[0157] It is easy to understand that this application, without relying on complex models, uses auditable and interpretable rules to map business value into the revenue per unit of resources, providing a monotonous and stable input for cut point localization.

[0158] S2.2.3: Using the difference between the shadow prices corresponding to the first communication link and the second communication link as the liquidation threshold, the transmission overhead of the data unit is accumulated sequentially from the beginning of the priority sequence until the accumulated value is less than the available service budget, and the marginal utility density of the current data unit is greater than or equal to the liquidation threshold, and the current data unit is taken as the liquidation cut point.

[0159] Specifically, the liquidation threshold represents the additional cost per unit required to upgrade a data unit from the second communication link to the first communication link, and it is directly related to the shadow price.

[0160] Understandably, the process of locating a cut point involves finding the last still worthwhile position in the priority sequence that doesn't exceed the budget. Operationally, scanning and accumulation begin from the beginning of the sequence: as long as the marginal utility density of the current data unit is not lower than the liquidation threshold, and the accumulated expected capacity is still within the available service budget, the data unit is included in the first communication link; when any condition is no longer met, the current data unit is identified as the liquidation cut point.

[0161] In some optional implementations, to improve the stability and operability of clearing cut point positioning, two layers of hysteresis bands are set around the clearing cut point:

[0162] Boundary terms falling into the inner hysteresis band adopt a dual-link sharding parallel strategy, that is, the second communication link is used to send the digest or low-cost parameters first, and the key residuals are then added in the parasitic activity window of the first communication link.

[0163] Boundary terms falling into the outer hysteresis band maintain the previous allocation results and are only adjusted when two consecutive forecast time windows show a liquidation threshold or a budget shift in the same direction.

[0164] For delayed or out-of-order entries, insertion is performed using a time watermark and backfilling mechanism without disrupting the established transaction order. In cases of significant dictionary consistency gaps, dictionary differences are inserted as virtual data units before the clearing cut point, prioritizing the use of a small portion of the budget in the first communication link to ensure that the sparse parameters of the subsequent second communication link can still be decoded. For windows where the budget is heavily occupied by upstream rigid tasks, degradation is allowed for several supporting entries before the cut point, and compensation placeholders are recorded, indicating that backfilling will be performed in the next active window by merging small packets or using micro-fragments.

[0165] It is easy to understand that in the aforementioned optional specific implementation, the clearing cut point position will not swing back and forth between the two links due to instantaneous jitter, the number of cellular wake-up times and handshake frequency are effectively controlled, the second communication link maintains a continuous and decodeable carrying capacity for a long time, and the overall timeliness, energy consumption and cost are comprehensively balanced.

[0166] In some alternative implementations, geometric curve modeling can be used to locate the clearing cut point.

[0167] Specifically, within a preset decision time window, the expected transmission overhead of each data unit under the second communication link is mapped to the horizontal axis, and the marginal utility of its unit transmission overhead for settlement correctness, control timeliness, or anomaly identification is mapped to the vertical axis. These components are then connected sequentially according to priority to form a monotonic broken line.

[0168] Based on the monotonic polygonal line, shape regularization is performed on the monotonic polygonal line to obtain the upper envelope curve, and geometric features such as inflection points, curvature changes and slope transitions are extracted from the curve to characterize the trend of service value changing with transmission cost.

[0169] Furthermore, based on parameters such as latency, packet loss, energy consumption, congestion, and wake-up / authentication overhead amortization of the first and second communication links within the time window, a price band reference line is generated to characterize the difference in the overall cost of the two links. By comparing the upper envelope curve and the price band reference line, the location of the settlement cut point is determined within the interval where the two first form a stable intersection, combined with the service budget projection of the first communication link within the time window.

[0170] Next, we will further elaborate on the technical content of the sparse strategy in the method of this application.

[0171] In one example, generating the encoding information of the data unit using a sparse strategy includes:

[0172] S3.1: Perform type identification on the data unit, and select a target dictionary from the preset dictionary set according to the identification result, wherein the target field includes version identifier and fingerprint verification information;

[0173] Specifically, type recognition maps data units to dictionary families that match their statistical structure and event semantics, avoiding reconstruction bias and sparsity failures caused by processing heterogeneous data with a single dictionary. The recognition process first performs initial screening based on message headers and field semantics, extracting business categories (periodic meter reading, event alarms, cost control receipts, operation logs, load waveforms, etc.) and transaction stage information; then, it extracts lightweight structural features from the original or denoised sequences, including periodic indicators, trend stability, degree of local mutation, energy concentration, neighborhood similarity, and missing distribution; based on this judgment, it selects a target dictionary from a preset dictionary set, carrying version identifiers and fingerprint verification information to ensure subsequent end-to-end consistency and traceability.

[0174] In some optional implementations, the preset dictionary set adopts a multi-layered organization of "family-subclass-subspace"; for example, the periodic load dictionary family contains subclasses subdivided by season / time period, and the sudden event dictionary family contains subclasses refined by device type and alarm level; each subclass has several subspaces to adapt to different sampling granularities and window lengths. The type identification output includes: target dictionary family and subclass number, subspace number, version identifier, fingerprint digest, confidence score, and candidate dictionary pointers. When the confidence score is insufficient or there is a feature conflict, a conservative strategy is adopted, prioritizing the selection of the base subclass with higher tolerance, and explicitly marking it in the encoding header so that the receiver can trigger differential clarification when necessary.

[0175] S3.2: Perform sparse coding and local reconstruction on the data unit according to the target dictionary to obtain a sparse location set and quantization coefficients, and calculate the error estimate based on the local reconstruction result;

[0176] Specifically, sparse coding uses a target dictionary as a constraint to find a sparse representation that can describe data units with the fewest parameters, while controlling quantization error and reconstruction bias. Before encoding, normalization is performed (detrending, scale alignment, and mild outlier suppression), and the data is divided into blocks aligned with dictionary atoms. Then, a stepwise sparse solution process is executed: candidate atoms with the highest correlation to the current residual and not strongly correlated with the selected atoms in the support set are selected first, until the set sparsity or local error threshold is reached. Adaptive quantization is performed on the coefficients of the selected atoms, and the quantization step size is automatically set according to the coefficient distribution and target precision to achieve a balance between code rate and error. After completion, reconstruction is performed locally using the same target dictionary to obtain a reconstruction sequence for self-checking.

[0177] In this embodiment, the error estimate considers not only numerical deviation but also structural consistency. Numerical deviation can be comprehensively evaluated through dimensions such as residual energy after reconstruction, deviation of key statistics, and the degree of recovery of peaks or plateaus. Structural consistency is evaluated through indicators such as the stability of the support set, whether the distribution of residual energy across frequency bands is abnormally concentrated, and the degree of overlap with historical support trajectories. When the reconstruction result presents a potential risk of misjudgment at key elements, even if the overall numerical deviation is not large, the error estimate is still increased according to the risk priority to promote the subsequent triggering of remedial mechanisms. This process is completed at the edge, and can predict whether differential, residual, or link switching is needed without relying on feedback from the receiver, thereby shortening the closed-loop delay.

[0178] S3.3: If the error estimate is greater than the preset error threshold, update the preset dictionary set according to the error estimate to obtain a dictionary difference update package;

[0179] Specifically, when the error estimate shows that the existing target dictionary cannot reconstruct the current data unit at low cost, or when the support set continues to drift within a short window, it indicates that the data statistical structure has shifted. In this case, the dictionary needs to be fine-tuned without transmitting the complete dictionary. The differential update package describes the transition from the current version to the target version in the form of a sequence of atomic operations, including adding atoms, eliminating aging atoms, adjusting the weights or shifting the center of gravity of existing atoms, changing the mapping order of atoms in the subspace, and fine-tuning the quantization table in intervals. Each atomic operation carries the necessary parameters and minimum verification information, and generates an integrity summary and parent version fingerprint at the segment level for the receiving end to apply incrementally by segment and for cross-version verification.

[0180] S3.4: Generate an encoding header corresponding to the data unit based on the dictionary difference update packet, the sparse location set, and the quantization coefficients, and send a transmission message including the encoding header to the receiving end through the second communication link.

[0181] Next, combined Figure 7 This application introduces another method for adaptive switching of dual-mode communication in smart meters, as provided in its embodiments. Figure 7 The method shown is applied to the receiving end of a communication system, which includes a first communication link and a second communication link, wherein the transmission rate of the second communication link is less than that of the first communication link. The communication system also includes a transmitting end, which is communicatively connected to the receiving end through the first and second communication links. The transmitting end and the receiving end are configured with the same preset dictionary set. The specific steps of the method are as follows:

[0182] A1: Receive the first type of transmission message from the first communication link and the second type of transmission message from the second communication link, wherein the first type of transmission message is an original message and the second type of transmission message is a sparsely encoded message;

[0183] A2: Establish the correspondence between the second type of transmission messages and the first type of transmission messages;

[0184] In this embodiment, the first type of message refers to the original message transmitted through the first communication link, and the second type of message refers to the sparsely encoded message transmitted through the second communication link. Since the two types of messages differ in link latency and arrival order, the receiving end needs to buffer them separately and retain transaction indexes and time raster indexes in the cross-link ledger to ensure asynchronous alignment later. This is done to avoid situations where high-speed messages arrive first but lack completion, or low-speed messages arrive late, resulting in lost context. By reserving double buffers and an asynchronous index table, it can be guaranteed that messages, regardless of their arrival order, can be included in the unified processing link.

[0185] A3: For the second type of transmission message, determine whether it carries indication information of dictionary differential update packet;

[0186] In this embodiment, the receiving end determines whether a differential update packet is attached by checking the version identifier and indicator bit in the packet header. If the version fingerprint is found to be inconsistent with the local dictionary, the differential update processing logic is activated. This ensures that the dictionary used during sparse decoding is synchronized with the sending end; otherwise, the sparse parameters on low-speed links may not be correctly restored. By identifying and triggering the update at the receiving end in advance, decoding failures and retransmissions can be effectively reduced.

[0187] A4: If so, update the preset dictionary set according to the dictionary difference update package to obtain the updated dictionary set;

[0188] In this embodiment, the receiving end performs integrity verification and parent version fingerprint verification on the differential packets. After confirming that they have not been tampered with, it applies atomic operations, including adding atoms, deleting inefficient atoms, fine-tuning parameters, or adjusting the quantization table. The updated dictionary set replaces the old version and generates a new version number and checksum. The receiving end can quickly obtain a dictionary state consistent with the sending end, ensuring that all subsequent sparse packets can be correctly parsed, while also avoiding the waste of link resources due to frequent transmission of the complete dictionary.

[0189] A5: Parse the second type of transmission message according to the updated dictionary set to obtain the corresponding data unit reconstructed message;

[0190] In this embodiment, the receiving end uses the updated dictionary to decode the sparse location set and quantization coefficients, and performs reconstruction to obtain a reconstructed data unit message. During this process, the receiving end performs residual consistency verification and statistical feature comparison on the reconstruction result to ensure that the reconstructed message and the first type of message can form a logical closed loop. If an error exceeding the limit or a data gap is detected, the receiving end marks the message as needing compensation and requests residual compensation through uplink feedback. Even if there is jitter or data loss in the low-speed link, effective reconstruction can still be completed in most cases.

[0191] A6: Combine the data unit reconstructed message and the first type of transmission message to obtain the final message;

[0192] In this embodiment, the receiving end synthesizes the reconstructed message and the corresponding original message at the data unit level based on the cross-link association ledger. The specific synthesis strategy includes: if the reconstructed message only supplements the digest, the first type of message is used as the main component, with sparse details inserted; if the reconstructed message provides critical residuals, the reconstructed message is used as the main component, correcting statistical errors in the first type of message. The synthesized final message possesses integrity and consistency and can be directly used for settlement, control, or archiving. Through this synthesis mechanism, the receiving end effectively utilizes the complementary characteristics of the two links: the high-speed link ensures timeliness, and the low-speed link provides accuracy, thus balancing stability and economy.

[0193] In one example, the specific steps for A2 are as follows:

[0194] A2.1: Construct a cross-link association ledger at the receiving end, wherein the cross-link association ledger is partitioned and managed according to the time grid of the first type of transmission message;

[0195] A2.2: For the first type of transmission message, extract the alignment binding code for cross-link alignment, and store the first type of transmission message in the cross-link association ledger according to the alignment binding code, wherein the alignment binding code is obtained by combining the transaction sequence number, time index and semantic anchor set calculated based on the original message of the first type of transmission message;

[0196] A2.3: For the second type of transmission message, extract the alignment candidate binding code for cross-link alignment, wherein the alignment candidate binding code is obtained by combining the transaction sequence number, time index and sparse support digest of the second type of transmission message;

[0197] A2.3: Based on the delay distribution of the second communication link, determine the time dilation window that matches the alignment candidate binding code, and retrieve the cross-link association ledger according to the time dilation window to obtain multiple candidate grids, wherein the candidate grids include at least one type of first-class transmission message;

[0198] A2.4: Determine the first type of transmission message corresponding to the second type of transmission message based on the alignment candidate binding code and the corresponding alignment binding code in the candidate grid.

[0199] The step of updating the preset dictionary set according to the dictionary difference update package includes:

[0200] The dictionary differential update package is subjected to integrity and source authentication, and the target version identifier, parent version fingerprint, segment index and atomic operation sequence are parsed.

[0201] The consistency of the preset dictionary set is verified based on the parent version fingerprint. If there is a discrepancy, the preset dictionary set is updated. The update includes: adding, deleting, replacing, rearranging and mapping, adjusting weights, merging subspaces and splitting subspaces of the preset dictionary set according to the atomic operation sequence and segment index, and gradually generating a verification hash. The version identifier of the preset dictionary set is modified according to the target version identifier.

[0202] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A dual-mode communication adaptive switching method for smart meters, applied to the transmitting end of a communication system, characterized in that, The communication system includes a first communication link and a second communication link, wherein the transmission rate of the second communication link is less than that of the first communication link. The communication system also includes a receiving end, which is communicatively connected to the sending end via the first and second communication links. The sending end and the receiving end are configured with the same preset dictionary set. The method includes: The data to be sent is divided into multiple data units, each of which includes at least one message. Obtain the link status parameters of the first and second communication links, and determine the target communication link corresponding to each data unit based on the priority of the data unit. Determining the target communication link includes: The priorities of multiple data units are aggregated to obtain a priority sequence. Based on the link status parameters, the clearing cut point in the priority sequence is located. Data units in the priority sequence that are before the clearing cut point are assigned to the first communication link, and data units in the priority sequence that are at or after the clearing cut point are assigned to the second communication link. If the target communication link corresponding to the data unit is the second communication link, the encoding information of the data unit is generated through a sparse strategy, and the encoding information is forwarded through the second communication link; The liquidation cut points in the location priority sequence include: Within a preset prediction time window, the shadow prices corresponding to the first communication link and the second communication link are calculated based on the link status parameters, and the available service budget of the first communication link in the prediction time window is determined. For each data unit in the priority sequence, calculate the marginal utility density of the data unit under unit transmission overhead; Using the difference between the shadow prices corresponding to the first communication link and the second communication link as the liquidation threshold, the transmission overhead of the data unit is accumulated sequentially from the beginning of the priority sequence until the accumulated value is less than the available service budget and the marginal utility density of the current data unit is greater than or equal to the liquidation threshold, and the current data unit is taken as the liquidation cut point. The step of generating the encoding information of the data unit using a sparse strategy includes: The data unit is type-identified, and a target dictionary is selected from the preset dictionary set according to the identification result, wherein the target dictionary includes version identifier and fingerprint verification information; The data unit is sparsely encoded and reconstructed locally according to the target dictionary to obtain a sparse location set and quantization coefficients, and the error estimate is calculated based on the local reconstruction result. If the error estimate is greater than the preset error threshold, the preset dictionary set is updated according to the error estimate to obtain a dictionary difference update package; Generate an encoding header corresponding to the data unit based on the dictionary differential update packet, the sparse location set, and the quantization coefficients, and send a transmission message including the encoding header to the receiving end through the second communication link.

2. The adaptive switching method for dual-mode communication in a smart meter according to claim 1, characterized in that, If the target communication link corresponding to the data unit is the first communication link, the corresponding data unit is sent to the receiving end through the first communication link.

3. The adaptive switching method for dual-mode communication in a smart meter according to claim 1, characterized in that, The link status parameters include at least one of latency, packet loss rate, bandwidth utilization, and energy consumption indicators.

4. The adaptive switching method for dual-mode communication in a smart meter according to claim 1, characterized in that, The priority is calculated in the following ways: The messages in the data unit are characterized to generate a summary layer for representing business semantics and multiple residual layers with increasing information granularity. The summary layer is prioritized according to a preset business semantic strategy library to obtain the initial priority of the data unit; The effect index of each residual layer is calculated, and the effect index is synthesized into an effect score through a preset mapping function. The effect index includes at least one of the following: the marginal information gain of the residual, the substitutability index with historical samples, the remaining time margin, and the estimated transmission energy consumption. The priority of a data unit is obtained based on the initial priority and the effect score, combined with a preset grading threshold.

5. The adaptive switching method for dual-mode communication in a smart meter according to claim 1, characterized in that, Determining the target communication link corresponding to each data unit includes: The priorities of multiple data units are aggregated to obtain a priority sequence, wherein the priority sequence is sorted in descending order according to the priority of the data units; Based on the priority sequence and in conjunction with the link state parameters of the first and second communication links, the clearing cut point in the priority sequence is located. Data units in the priority sequence located before the clearing cut point are allocated to the first communication link, while data units in the priority sequence located after the clearing cut point and data units corresponding to the clearing cut point are allocated to the second communication link.

6. A dual-mode communication adaptive switching method for smart meters, applied to the receiving end of a communication system, characterized in that, The communication system includes a first communication link and a second communication link, wherein the transmission rate of the second communication link is less than that of the first communication link. The communication system also includes a transmitting end, which is communicatively connected to the receiving end via the first and second communication links. The transmitting end and the receiving end are configured with the same preset dictionary set. The method includes: Receive a first type of transmission message from the first communication link and a second type of transmission message from the second communication link, wherein the first type of transmission message is an original message and the second type of transmission message is a sparsely coded message; For the second type of transmission message, determine whether it carries indication information of dictionary differential update packet; If so, the preset dictionary set is updated according to the dictionary difference update package to obtain the updated dictionary set; The second type of transmission message is parsed according to the updated dictionary set to obtain the corresponding data unit reconstructed message; The data unit reconstructed message and the first type of transmission message are combined to obtain the final message; The method further includes establishing a correspondence between the second type of transmission messages and the first type of transmission messages, specifically including: At the receiving end, a cross-link association ledger is constructed, wherein the cross-link association ledger is partitioned and managed according to the time index of the first type of transmission message; For the first type of transmission message, an alignment binding code for cross-link alignment is extracted, and the first type of transmission message is stored in the cross-link association ledger according to the alignment binding code, wherein the alignment binding code is obtained by combining the transaction sequence number, time index and semantic anchor set calculated based on the original message of the first type of transmission message; For the second type of transmission message, an alignment candidate binding code for cross-link alignment is extracted, wherein the alignment candidate binding code is obtained by combining the transaction sequence number, time index, and sparse support digest of the second type of transmission message; Based on the delay distribution of the second communication link, a time dilation window matching the alignment candidate binding code is determined, and the cross-link association ledger is retrieved based on the time dilation window to obtain multiple candidate grids, wherein the candidate grids include at least one type of first-class transmission message. Based on the alignment candidate binding code and the corresponding alignment binding code in the candidate grid, the first type of transmission message corresponding to the second type of transmission message is determined.

7. The adaptive switching method for dual-mode communication in a smart meter according to claim 6, characterized in that, The step of updating the preset dictionary set according to the dictionary difference update package includes: The dictionary differential update package is subjected to integrity and source authentication, and the target version identifier, parent version fingerprint, segment index and atomic operation sequence are parsed. The consistency of the preset dictionary set is verified based on the parent version fingerprint. If there is a discrepancy, the preset dictionary set is updated. The update includes: adding, deleting, replacing, rearranging and mapping, adjusting weights, merging subspaces and splitting subspaces of the preset dictionary set according to the atomic operation sequence and segment index, and gradually generating a verification hash. The version identifier of the preset dictionary set is modified according to the target version identifier.

Images