A SMS channel scheduling method based on the Crush algorithm
By using the Crush algorithm and multi-dimensional real-time weighted scheduling, combined with a state awareness mechanism that is timed and event-triggered, the problems of uneven load and slow fault response in SMS channel scheduling are solved, achieving high efficiency and high reliability in SMS sending and adapting to different business needs.
Patent Information
- Application Number
- CN202611600801.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-16
- Publication Date
- 2026-11-17
AI Technical Summary
Existing SMS channel scheduling methods suffer from uneven load distribution, poor adaptability, and slow fault response, resulting in low SMS sending efficiency and low success rate, making it difficult to guarantee business continuity, especially in high-concurrency request scenarios.
By employing the Crush algorithm combined with multi-dimensional real-time weighted scheduling logic and a state awareness mechanism triggered by timed events, we can achieve immediate isolation and rescheduling of faulty channels, dynamically adjust request distribution strategies, and optimize weight configuration and detection frequency.
It improves the overall efficiency and success rate of SMS sending, ensures the continuous operation of high-priority services, adapts to usage scenarios with different business volumes and quality requirements, and optimizes system resource utilization.
Smart Images

Figure CN123123829A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of communication technology, specifically to a method for scheduling SMS channels based on the Crush algorithm. Background Technology
[0002] SMS, as a widely used and highly effective communication method, is widely applied in various business scenarios such as financial identity verification, government notifications, and enterprise service announcements. With the continuous growth in SMS volume across various services, the parallel deployment of multiple SMS channels has become a common way to improve overall delivery capacity. The rationality of the channel scheduling method directly determines the efficiency and quality of SMS delivery. Currently, mainstream SMS channel scheduling technologies mainly include three implementation approaches: polling scheduling, random scheduling, and simple weighted scheduling. Each scheduling method has its own emphasis on implementation complexity and scheduling effectiveness.
[0003] Polling scheduling allocates SMS sending requests sequentially in a fixed order. While the logic is simple, it fails to consider the varying capacity and real-time operational status of different channels, easily leading to high-load channels continuously receiving requests, causing delays and increased failure rates. Random scheduling distributes requests randomly, failing to accurately match request load with channel capacity. This often results in some channels being overloaded while others are idle, wasting channel resources. Simple weighted scheduling, while introducing channel attribute dimensions to set allocation ratios, relies on statically configured weight parameters that cannot dynamically update with channel network fluctuations and load changes. When channel operational status shifts, the scheduling strategy cannot adapt in a timely manner. Furthermore, existing scheduling methods have long channel fault detection and switching links, hindering rapid request migration when a channel fails. In high-concurrency scenarios, scheduling response delays are also prone to occur, making it difficult to guarantee business continuity. Summary of the Invention
[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a text message channel scheduling method based on the Crush algorithm. This method combines the Crush algorithm with multi-dimensional real-time weighted scheduling logic, a state awareness mechanism that combines timed and event-triggered mechanisms, and a faulty channel immediate isolation and rescheduling process. This solves the problems of uneven load distribution, poor adaptability, and slow fault response in traditional scheduling methods. The method can dynamically adjust the request distribution strategy according to the real-time operating status of the channel, quickly identify and handle channel faults, and periodically optimize weight configuration and detection frequency based on historical scheduling data. While ensuring text message sending efficiency and delivery quality, it improves system resource utilization and adapts to text message sending scenarios with different business volumes and quality requirements.
[0005] To solve the above-mentioned technical problems, this invention provides the following technical solution: a method for scheduling SMS channels based on the Crush algorithm, the specific steps of which are as follows: S1. Collect basic attribute information of all available SMS channels. After data cleaning to remove invalid and duplicate data, assign a unique identifier to each SMS channel, build an SMS channel information database, and configure the mapping rule parameters and weight calculation factors of the Crush algorithm. S2. A mechanism combining timed detection and event-triggered detection is adopted to collect real-time operating status data of each SMS channel and update it synchronously to the SMS channel information database. S3. Retrieve basic attribute information and real-time operating status data from the SMS channel information database, and calculate the real-time weight of each SMS channel by combining it with the preset weight calculation factor; reset the real-time weight of SMS channels that are determined to be faulty to 0 and exclude them from the scheduling scope. S4. Receive SMS sending requests and extract request features. Filter available channels according to the operator to which the recipient's number belongs. Input the identifier and real-time weight of the available channels into the Crush algorithm to generate a target channel list sorted by priority. Select the target channel based on the priority of the SMS sending request and execute SMS sending. S5. During the SMS sending process, monitor the sending status of the target channel in real time. If the channel is found to be faulty, update the real-time weight of the corresponding channel to 0 and remove it from the set of available channels. Then, re-execute S4 to complete the rescheduling of the SMS sending request. S6. Analyze and schedule historical data according to a preset cycle, and dynamically adjust the value of the weight calculation factor and the time interval of the timed detection.
[0006] Furthermore, in S1, the basic attribute information of the SMS channel includes channel identifier, rated bandwidth, maximum concurrency, historical sending success rate, initial network latency, and carrier type; the weight calculation factor includes bandwidth weight factor. Concurrency weighting factor Success rate weighting factor With delay weighting factor And satisfy This clarifies the dimensional composition and weight constraints of scheduling decisions, ensuring that the proportion of influence of each dimension matches the actual needs of the business scenario.
[0007] Furthermore, in S2, the time interval for timed detection ranges from 1 minute to 5 minutes; the triggering conditions for event-triggered detection include the channel transmission failure rate exceeding a preset threshold and the network latency increase exceeding a preset threshold; the timed detection obtains the response latency by sending a test SMS to the channel, and linearly lowers the channel success rate value according to the multiple by which the latency exceeds the preset threshold, with the lowering magnitude being positively correlated with the multiple. Through dual-mode monitoring, the real-time nature of state awareness and system resource overhead are taken into account, and abnormal fluctuations in channel operation are captured in a timely manner.
[0008] Furthermore, in S3, the formula for calculating the real-time weight W is: Where B is the currently available bandwidth, which is a non-negative value of the difference between the rated bandwidth and the bandwidth occupied by the current number of concurrent users; C is the currently available number of concurrent users, which is the difference between the maximum number of concurrent users and the current number of concurrent users; S is the current transmission success rate; and D is the reciprocal of the current network latency. Through multi-dimensional weighted calculation, the actual carrying capacity and transmission quality of the channel are comprehensively reflected, providing a quantitative basis for scheduling decisions.
[0009] Furthermore, the currently available bandwidth B is calculated as follows: ;in, Where N is the channel's rated bandwidth and N is the current concurrent users. The bandwidth usage per SMS message is preset based on the average length of a single SMS message, avoiding logical contradictions such as negative values in bandwidth calculations and accurately reflecting the actual availability of remaining bandwidth resources in the channel.
[0010] Furthermore, in S4, the mapping rules of the Crush algorithm support filtering channel groups based on one or more of the following conditions: sender number, receiver number, SMS content keywords, SMS length, channel billing rate, and channel success rate. The priority of the target channel list is positively correlated with the real-time weight of the channel; the higher the weight, the higher the priority ranking. This supports flexible adaptation to the channel filtering needs of different services and improves the scenario adaptability of the scheduling method.
[0011] Furthermore, in S4, the rule for selecting the target channel based on request priority is as follows: high-priority requests select the channel ranked first in the target channel list; medium-priority requests randomly select from the channels ranked second to third in the target channel list; and low-priority requests randomly select from the channels ranked fourth to fifth in the target channel list. This achieves a corresponding match between different priority services and channel quality, ensuring the transmission quality of high-priority services while also taking into account overall load balancing.
[0012] Furthermore, in S5, the criteria for determining a channel failure are: three to five consecutive SMS sending failures, or detection of a channel network interruption or response timeout; after the failure is determined, the failure status is immediately marked in the SMS channel information database, and the real-time weight of the corresponding channel is reset to 0, clarifying the failure determination criteria and processing actions, ensuring that the faulty channel is removed from the scheduling range in a timely manner, and avoiding continuous allocation of requests that lead to batch sending failures.
[0013] Furthermore, in S6, the scheduling history data includes the SMS sending volume, sending success rate, average sending delay, and number of fault occurrences for each channel. The adjustment rule for the weight calculation factor is as follows: for SMS channels with a sending success rate higher than a preset threshold and no fault records within a preset statistical period, the corresponding success rate weight factor is increased; for SMS channels with fault records within a preset statistical period and a sending success rate lower than a preset threshold, the corresponding concurrency weight factor is decreased. The adjustment rule for the timed detection interval is as follows: based on the time period distribution of business volume in the scheduling history data, the time interval for timed detection is shortened during peak periods when the business sending volume is higher than a preset threshold, and the time interval for timed detection is extended during trough periods when the business sending volume is lower than a preset threshold. This achieves iterative optimization of the scheduling strategy, ensuring that the weight configuration and detection frequency match the actual business operation, and optimizing system resource consumption while ensuring scheduling effectiveness.
[0014] Compared with existing technologies, this SMS channel scheduling method based on the Crush algorithm has the following advantages: I. This invention uses the Crush algorithm combined with a multi-dimensional real-time weighted scheduling mechanism to incorporate available channel bandwidth, remaining concurrency, sending success rate, and network latency into the scheduling decision dimensions. It can dynamically adjust the request distribution strategy according to the real-time operating status of the channel, prioritizing the allocation of SMS sending requests to channels with sufficient resources and better operating status. This solves the problems of uneven load and poor adaptability of traditional static scheduling methods, and improves the overall efficiency and success rate of SMS sending.
[0015] Second, this invention uses a channel status awareness mechanism that combines timed detection and event-triggered detection, along with a faulty channel immediate isolation and rescheduling process, to quickly identify channel faults and seamlessly switch sending requests, avoiding SMS sending interruptions and ensuring the continuous operation of high-priority services.
[0016] Third, this invention uses a periodic scheduling data analysis mechanism to dynamically adjust the weight calculation factor and the timed detection frequency. It can match the corresponding monitoring intensity according to the time period changes of business volume, optimize system resource consumption while ensuring scheduling accuracy, and adapt to different business volume levels and different quality requirements in different application scenarios.
[0017] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0019] Figure 1 This is a schematic diagram of the overall process of the SMS channel scheduling method based on the Crush algorithm; Figure 2 A schematic diagram of the real-time weight calculation process for SMS channels; Figure 3 This is a schematic diagram of the SMS channel fault detection and rescheduling process. Detailed Implementation
[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.
[0021] This embodiment applies to an SMS sending system for verification code and notification services. This system handles the sending of various SMS messages, including those for identity verification and business notifications, and requires high timeliness and delivery rates. The system connects to 10 available SMS channels, belonging to three major telecom operators. The rated parameters of each channel are collected and entered during the connection process. For example... Figure 1 As shown, the overall execution flow of this method proceeds sequentially according to the logic of initialization configuration, status monitoring, weight calculation, request scheduling, fault handling, and strategy optimization.
[0022] After system startup, the first step is to execute the channel information collection and initialization process. Basic attribute information for all available SMS channels is collected sequentially, including rated bandwidth, maximum concurrency, historical success rate, initial network latency, and carrier type. After collection, data cleaning is performed, removing invalid data without valid carrier identifiers or with rated parameters of zero. For multiple duplicate records of the same channel, the latest record is retained. After cleaning, a unique integer identifier is assigned to each SMS channel, and the basic attribute information is stored in a one-to-one correspondence with the channel identifier, constructing an SMS channel information database. The database includes a channel basic information table and a real-time status table, storing static attributes and dynamic operational data respectively. Subsequently, the parameters for the Crush algorithm are initialized. The mapping function uses a standard controllable hash mapping function, and the replica count is set to 1. The initial values of the weight calculation factors are set according to the priority of business scenario requirements. Since SMS sending success rate has the greatest impact on business in this scenario, the success rate weight factor is set to the highest proportion, followed by bandwidth and concurrency weight factors, and then network latency weight factor, with the sum of the values of the four weight factors being 1. The bandwidth usage per SMS message is set based on the average length of a single SMS message in the business scenario. In this scenario, the length of a single SMS message is generally within 70 characters, corresponding to the bandwidth usage of a standard SMS message. This serves as the benchmark for calculating the bandwidth usage per unit. The basic parameters of each SMS channel are shown in Table 1.
[0023] Table 1:
[0024] After initial configuration, the system initiates a real-time channel status monitoring process, employing a combination of timed detection and event-triggered detection. Timed detection is executed via the system's built-in timed task scheduler, with an initial time interval set to 2 minutes. At each detection time, the system sends a test SMS to all online channels, records the sending timestamp, waits for a status report from the channel, records the return timestamp, and calculates the difference between the two timestamps to obtain the end-to-end response latency. If the response latency exceeds a preset time threshold, the success rate of the corresponding channel is linearly reduced according to the multiple of the excess. The reduction is positively correlated with the multiple of the excess; exceeding the threshold by 1 unit results in a 1-unit reduction, exceeding by 2 units results in a 2-unit reduction, and so on. Event-triggered detection is driven by channel operation anomalies, with a statistical window set to 5 minutes. Every 5 minutes, the system compiles downlink SMS receipts for each channel and records the final status of each SMS. When the transmission failure rate of a channel exceeds a preset threshold within the statistical window, or when the network latency increases by more than a preset threshold within a short period, a specific detection for that channel is immediately triggered. The failure rate is calculated by summing the number of SMS messages with an undelivered status and the number of SMS messages that did not return a status report, and dividing that number by the total number of SMS messages sent within the statistical window. After the special testing is completed, the latest status data is synchronized and updated to the real-time status table in the SMS channel information database. At a given monitoring moment, the real-time operating data for each channel is shown in Table 2.
[0025] Table 2:
[0026] After the status data is updated, the system performs real-time channel weight calculation. For example... Figure 2 As shown, the weight calculation sequentially completes the parameter calculation for each dimension, weighted summation, and fault channel weight correction, ultimately outputting real-time weight values that can be directly used for scheduling decisions. The real-time weight is obtained by multiplying the parameters of the four dimensions by their corresponding weight factors and then summing the results. The calculation formula is as follows: Where W is the real-time channel weight and B is the currently available bandwidth. Where C is the bandwidth weighting factor and C is the current available concurrency. Here, S is the concurrency weighting factor, and S is the current transmission success rate. The success rate is a weighting factor, and D is the reciprocal of the current network latency. This is the delay weighting factor.
[0027] The available bandwidth is calculated as follows: In the formula Where N is the channel's rated bandwidth and N is the current concurrent users. This represents the bandwidth usage per SMS message. The calculation result is taken as a non-negative value; when the result is less than zero, the available bandwidth is set to 0. This is because when the current traffic volume of the channel exceeds its rated bandwidth capacity, there is no remaining bandwidth resource to handle new sending requests, therefore the contribution of the bandwidth dimension is zero. It should be noted that rated bandwidth represents the channel's SMS processing capacity per unit time, while maximum concurrency represents the number of simultaneous connection requests the channel can handle. These are different operational parameters, corresponding to processing speed and capacity capabilities, respectively. Available concurrency is the channel's maximum concurrency minus the current concurrency, representing the remaining connection carrying capacity. The current sending success rate is a value obtained from real-time monitoring and is used in the calculation as a decimal. The latency dimension is the reciprocal of the current network latency; the lower the latency, the higher the contribution of this dimension. After calculation, if the channel is marked as faulty, its real-time weight is directly set to 0, and it does not participate in subsequent scheduling and allocation.
[0028] In this embodiment, the real-time weight of each channel is calculated one by one according to the above formula. Taking channel 2 as an example, the rated bandwidth is 120 messages / second, the current concurrent number is 350 messages, and the bandwidth occupied by a single SMS message is 1 message / second. The calculated bandwidth usage exceeds the rated value, so the available bandwidth is 0. The maximum concurrent number is 600 messages, and the current concurrent number is 350 messages, so the available concurrent number is 250 messages. The current sending success rate is 99%, which is converted to a decimal of 0.99. The current network latency is 44ms, and the reciprocal of the latency is approximately 0.0227. After substituting into the formula, the real-time weight of this channel is approximately 62.85. The remaining channels are calculated according to the same logic, and the final real-time weight values of all channels are shown in Table 3. The results are synchronously updated to the database.
[0029] Table 3:
[0030] When the system receives a new SMS sending request, it executes an SMS scheduling process based on the Crush algorithm. First, it extracts the characteristic information of the SMS sending request, including the recipient's number, SMS content length, and sending priority. Sending priorities are pre-defined according to service type: verification SMS messages have high priority, general notification SMS messages have medium priority, and promotional SMS messages have low priority. Then, it matches the recipient's number segment with the corresponding carrier type, filtering available channels from the SMS channel information database that match the carrier type and have a real-time weight greater than 0. The filtered channel identifier and its corresponding real-time weight are input into the Crush algorithm. The algorithm performs a hash operation on the recipient's number field of the request, mapping the hash value to a channel range divided by weight. Channels with higher weights have a larger mapping range and a higher probability of the request hitting the target. After mapping calculation, a target channel list is generated, sorted from highest to lowest priority, with higher-weight channels ranking higher in the list. Finally, based on the SMS request's sending priority, the corresponding target channel is selected from the target channel list. High-priority requests select the channel ranked 1st in the list to ensure optimal transmission quality; medium-priority requests randomly select one channel from the second to third ranked channels in the list to balance transmission quality and load balancing; low-priority requests randomly select one channel from the fourth to fifth ranked channels in the list to make full use of channel resources.
[0031] For example, when the system receives a high-priority verification SMS request, and the recipient's number belongs to the primary carrier, three matching available channels are obtained after filtering. The identifiers and weights of the three channels are input into the Crush algorithm, generating a target channel list ordered from highest to lowest priority: Channel 2, Channel 1, and Channel 3. Since the request is high-priority, Channel 2, ranked first, is selected as the sending channel, and the SMS request is distributed to this channel for transmission.
[0032] During SMS sending, the system continuously monitors the sending status of the target channel and records the sending result of each SMS message in real time. For example... Figure 3 As shown, fault detection and rescheduling follow a process of real-time monitoring, fault determination, channel isolation, and rescheduling to ensure uninterrupted transmission requests. When the number of consecutive transmission failures on the same channel reaches a preset threshold, or when abnormal conditions such as channel network interruption or response timeout are detected, the channel is determined to be faulty. The statistics for consecutive failures are based on different SMS messages processed consecutively by the channel; multiple retries of a single SMS message are not counted in the consecutive failure count. The fault determination threshold is set to 3 to 5 times; in this embodiment, it is set to 3 times.
[0033] Upon determining that a channel has failed, the system immediately performs a fault isolation operation. The channel is marked as faulty in the SMS channel information database, its real-time weight is updated to 0, and it is removed from the available channel set. The scheduling process is then re-executed, filtering available channels again based on the recipient's operator. The identifiers and weights of the remaining available channels are input into the Crush algorithm to generate a new list of target channels. Target channels are then reselected according to the original request's priority, and SMS requests are distributed to the new target channels to continue sending, ensuring uninterrupted SMS communication.
[0034] For example, during the aforementioned sending process, if Channel 2 experiences three consecutive SMS sending failures, the system determines that the channel has malfunctioned. The real-time weight of Channel 2 is immediately updated to 0 and marked as faulty. Available channels from the corresponding operator are then re-selected, leaving Channel 1 and Channel 3. The Crush algorithm is then applied to generate a new list of target channels, in the order of Channel 1 and Channel 3. Since the request still has high priority, Channel 1, ranked first, is selected as the alternative channel. The SMS request is redistributed, and the SMS sending is finally completed.
[0035] The system dynamically optimizes the scheduling strategy according to a preset period. In this embodiment, the statistical period is set to 12 hours. After each statistical period, the system extracts the historical scheduling data of all channels within that period, including the total transmission volume, average transmission success rate, average transmission delay, and number of failures for each channel, as shown in Table 4.
[0036] Table 4:
[0037] Based on statistical data, the weight calculation factors and timing detection intervals are adjusted. The adjustment rule for the weight calculation factors is as follows: for channels with a transmission success rate higher than a preset threshold and no fault records within the statistical period, the corresponding success rate weight factor is appropriately increased, and the corresponding bandwidth weight factor is appropriately decreased, keeping the sum of the four weight factors unchanged. For channels with fault records within the statistical period and a transmission success rate lower than a preset threshold, the corresponding concurrency weight factor is appropriately decreased, and the weight factors of other dimensions are adjusted accordingly, keeping the sum unchanged. In this embodiment, for the three channels with a success rate higher than 99% and no faults for two consecutive periods, their success rate weight factor is increased from 0.35 to 0.38, and the corresponding bandwidth weight factor is decreased by 0.03. For the channel with one fault and a low success rate, its concurrency weight factor is decreased from 0.25 to 0.22, and the other factors are adjusted accordingly to keep the sum of 1.
[0038] The adjustment of the timed detection interval is based on the time-period distribution of traffic volume. By statistically analyzing historical transmission volume, peak and off-peak periods are identified. For periods where the transmission volume accounts for more than 10% of the total daily volume, peak periods are defined, and the timed detection interval is shortened to improve the sensitivity of status monitoring. For off-peak periods with a lower transmission volume, the timed detection interval is appropriately extended to reduce unnecessary system resource consumption. In this embodiment, the transmission volume from 10:00 to 12:00 and from 18:00 to 20:00 accounts for 40% of the total daily volume and is defined as peak periods. The timed detection interval for these periods is shortened from 2 minutes to 1 minute. The detection interval remains at 2 minutes for all other periods.
[0039] This embodiment fully covers the entire process from initial configuration to dynamic optimization. The execution logic and parameter settings of each step are reproducible and can be directly deployed in similar SMS sending systems to achieve dynamic channel scheduling and rapid fault switching.
[0040] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.
Claims
1. A method for SMS channel scheduling based on the Crush algorithm, characterized in that, The method includes the following steps: S1. Collect basic attribute information of all available SMS channels. After data cleaning to remove invalid and duplicate data, assign a unique identifier to each SMS channel, build an SMS channel information database, and configure the mapping rule parameters and weight calculation factors of the Crush algorithm. S2. A mechanism combining timed detection and event-triggered detection is adopted to collect real-time operating status data of each SMS channel and update it synchronously to the SMS channel information database. S3. Retrieve basic attribute information and real-time operating status data from the SMS channel information database, and calculate the real-time weight of each SMS channel by combining it with the preset weight calculation factor; reset the real-time weight of SMS channels that are determined to be faulty to 0 and exclude them from the scheduling scope. S4. Receive SMS sending requests and extract request features. Filter available channels according to the operator to which the recipient's number belongs. Input the identifier and real-time weight of the available channels into the Crush algorithm to generate a target channel list sorted by priority. Select the target channel based on the priority of the SMS sending request and execute SMS sending. S5. During the SMS sending process, monitor the sending status of the target channel in real time. If the channel is found to be faulty, update the real-time weight of the corresponding channel to 0 and remove it from the set of available channels. Then, re-execute S4 to complete the rescheduling of the SMS sending request. S6. Analyze and schedule historical data according to a preset cycle, and dynamically adjust the value of the weight calculation factor and the time interval of the timed detection.
2. The Crush algorithm based SMS channel scheduling method as claimed in claim 1, wherein, In the S1, the basic attribute information of the short message channel includes channel identification, rated bandwidth, maximum concurrency, historical sending success rate, initial network delay and operator type; the weight calculation factor includes bandwidth weight factor , concurrency weight factor , success rate weight factor and delay weight factor , and satisfies .
3. The Crush algorithm based SMS channel scheduling method as claimed in claim 1, wherein, In S2, the time interval for timed detection ranges from 1 minute to 5 minutes; the triggering conditions for event-triggered detection include the channel transmission failure rate exceeding a preset threshold and the network latency increase exceeding a preset threshold. The system periodically detects response latency by sending test SMS messages to the channel. The success rate of the channel is linearly reduced based on the multiple by which the latency exceeds a preset threshold, with the reduction magnitude being positively correlated with the multiple by which the latency exceeds the threshold.
4. The Crush algorithm based SMS channel scheduling method as claimed in claim 2, wherein, The calculation formula of the real-time weight W in the S3 is: ; wherein, B is the current available bandwidth, taking the non-negative value of the difference between the rated bandwidth and the current concurrent bandwidth; C is the current available concurrency, taking the difference between the maximum concurrency and the current concurrency; S is the current sending success rate; and D is the inverse of the current network delay.
5. The Crush algorithm based SMS channel scheduling method as claimed in claim 4, wherein, The current available bandwidth B is calculated in the following manner: ; wherein, is the channel rated bandwidth, N is the current concurrency, is the unit short message bandwidth occupancy, which is preset according to the average length of a single short message.
6. The Crush algorithm based SMS channel scheduling method as claimed in claim 1, wherein, In S4, the mapping rules of the Crush algorithm support filtering channel groups by one or more of the following conditions: sender number, receiver number, SMS content keywords, SMS length, channel billing rate, and channel success rate; the priority of the target channel list is positively correlated with the real-time weight of the channel, and the higher the weight, the higher the priority ranking.
7. The Crush algorithm based SMS channel scheduling method as claimed in claim 1, wherein, In S4, the rule for selecting the target channel based on request priority is: for high-priority requests, the channel ranked first in the target channel list is selected. Medium-priority requests are randomly selected from the channels ranked 2nd to 3rd in the target channel list; low-priority requests are randomly selected from the channels ranked 4th to 5th in the target channel list.
8. The Crush algorithm based SMS channel scheduling method as claimed in claim 1, wherein, In S5, the criteria for determining a channel failure are: three to five consecutive SMS sending failures, or detection of a channel network interruption or response timeout; after the failure is determined, the failure status is immediately marked in the SMS channel information database, and the real-time weight of the corresponding channel is reset to 0.
9. The Crush algorithm based SMS channel scheduling method as claimed in claim 1 wherein, In S6, the scheduling history data includes the number of SMS messages sent, the success rate of sending, the average sending delay, and the number of failures for each channel; The adjustment rule of the weight calculation factor is: for the short message channel with the sending success rate higher than the preset threshold and no failure record in the preset statistical period, the success rate weight factor corresponding to the short message channel is adjusted upward; for the short message channel with the failure record and the sending success rate lower than the preset threshold in the preset statistical period, the concurrent number weight factor corresponding to the short message channel is adjusted downward; The adjustment rule of the timing detection interval is: according to the traffic time period distribution in the scheduling history data, the time interval of the timing detection is shortened in the peak period with the traffic sending amount higher than the preset threshold, and the time interval of the timing detection is lengthened in the valley period with the traffic sending amount lower than the preset threshold.
