Smart television system supporting remote storage management and control method

Abstract

The invention relates to the technical field of television storage management, in particular to an intelligent television system supporting remote storage management and a control method thereof.The intelligent television system comprises a user behavior portrait modeling unit, a multi-source collaborative decision-making unit and a remote configuration interface unit. Behavior portrait parameters are generated through a pre-trained long and short-term memory network, the content access probability is calculated according to the behavior portrait parameters, high-priority backup objects are marked, a multi-dimensional decision matrix is constructed in combination with storage and bandwidth, an optimal backup path is generated through a weighted entropy method and a Dijkstra algorithm, block incremental backup is performed, and breakpoint resuming is positioned through Hash comparison after transmission interruption. According to the method, the problems of backup resource waste and poor path adaptation are solved, and the remote storage management efficiency and reliability of the smart television are improved.

Description

Technical Field

[0001] This invention relates to the field of television storage management technology, and more specifically, to an intelligent television system and control method that supports remote storage management. Background Technology

[0002] TV storage management is an important technology. With the continuous enrichment of smart TV functions, the volume of movies, variety shows, and dramas watched by users continues to grow. The local storage capacity of smart TVs is gradually becoming insufficient to meet the long-term content retention needs. Remote storage has become an important supplement. The core role of this technology is to alleviate the pressure on the local storage of smart TVs through systematic remote storage management, avoiding the problem of users having to frequently delete content due to insufficient storage. It can also optimize content backup efficiency, ensuring that users' preferred content can be accessed at any time. Furthermore, it can balance storage costs, network overhead, and device reliability, promoting the upgrade of smart TVs from local storage dependence to local-remote collaborative storage, improving the long-term user experience, and widely adapting to smart TV devices in home entertainment scenarios. Existing remote storage management technologies for smart TVs face core problems in practical applications, including rigid backup strategies, poor path adaptability, and inefficient interruption handling. Regarding backup content selection, current technologies often use fixed rules to mark backup objects, failing to dynamically assess content value based on user viewing behavior. For example, they may back up content indiscriminately based on storage duration or type, neglecting user habits regarding following specific series or high-frequency viewing times. This results in a large amount of rarely accessed content being backed up, wasting remote storage space and home network bandwidth. Meanwhile, content that users truly need may be missed due to lack of marking. In terms of backup path selection, existing technologies often rely solely on network bandwidth or remote device distance, without comprehensively considering storage costs, network stability, and device reliability. For instance, they may choose only the path with the highest bandwidth while ignoring the high failure rate of the remote devices associated with that path. This can easily lead to backup interruptions or data corruption, or choosing low-cost remote storage but experiencing severe transmission delays due to excessive path hops. Existing technologies often use full retransmission for handling transmission interruptions, meaning the entire content file needs to be resent after an interruption, without segmenting or marking the content. If the interruption occurs late in the transmission process, most of the already transmitted data needs to be retransmitted, wasting bandwidth, significantly extending backup time, and further degrading the user experience. These problems combine to result in inefficient remote storage management, backup redundancy leading to wasted storage and bandwidth resources, improper paths increasing the risk of backup failures, and interrupted retransmissions extending backup cycles. Ultimately, remote storage cannot effectively supplement the backup process and may even be abandoned by users due to poor experience. To address this issue, we provide a smart TV system and control method that supports remote storage management. Summary of the Invention

[0003] The purpose of this invention is to provide a smart TV system and control method that supports remote storage management, so as to solve the problems mentioned in the background art.

[0004] Because of rigid backup strategies that fail to incorporate user behavior into backup content selection, resources are wasted or missed. Therefore, this case study uses a user behavior profiling modeling unit to collect viewing data and generate profile parameters. The multi-source collaborative decision-making unit calculates access probabilities based on these profile parameters and dynamically adjusts thresholds and marks high-priority objects, enabling accurate backups and saving resources.

[0005] Since full retransmission after a transmission interruption leads to wasted bandwidth and long processing time, this case study reduces redundant transmission and shortens backup time by generating hash identifiers for high-priority objects in blocks, comparing hashes to locate breakpoints during interruptions, resuming subsequent uploads, and verifying integrity.

[0006] To achieve the above objectives, a smart TV system supporting remote storage management is provided, including a user behavior profile modeling unit for real-time collection of user viewing behavior data and inputting it into a pre-trained long short-term memory network to generate user behavior profile parameters. A multi-source collaborative decision-making unit performs the following operations based on the user behavior profile parameters, remaining local storage space, real-time home network bandwidth, and remote storage device response latency: (a) Calculate the content access probability value based on user behavior profile parameters, and mark the content access probability value higher than the preset threshold as high priority backup objects; (b) Construct a multi-dimensional decision matrix that includes storage cost, network overhead, and device reliability based on the remaining local storage space, real-time bandwidth of the home network, and response latency of remote storage devices, and dynamically generate the optimal backup path by combining the multi-dimensional decision matrix with the weighted entropy method; (c) When the real-time bandwidth of the home network is detected to be higher than the preset network bandwidth for a fixed period of time and the remote storage device is idle, trigger the block incremental backup of the high-priority backup object; The remote configuration interface unit is used to provide an encrypted communication channel to the mobile terminal, receive the backup strategy weight parameters manually adjusted by the user, and synchronize them to the multi-source collaborative decision-making unit.

[0007] The second objective of this invention is to provide a method for implementing a smart TV system supporting remote storage management, including any of the above-mentioned features, comprising the following steps: S1. Capture single viewing duration, daily viewing time distribution, content type switching frequency, and continuous episode jump interval through a time series data collector. Input the behavioral data into a pre-trained long short-term memory network according to the time window sequence. The hidden layer of the network extracts time-related features and fuses them with the historical profile baseline to dynamically generate user behavior profile parameters. S2. Calculate the content access probability value based on user behavior profile parameters, mark high-priority backup objects in combination with dynamic baseline adjustment mechanism, and construct a multi-dimensional decision matrix based on local storage remaining space, home network real-time bandwidth and remote storage device response latency. Calculate information entropy weights through weighted entropy value method and drive Dijkstra optimization algorithm to generate the optimal backup path. S3. Divide high-priority backup objects into fixed-size data blocks and generate unique hash identifiers. When the real-time bandwidth of the home network is detected to be stable and the remote storage device is idle, transmit incremental data blocks in the local hash identifier library. If the transmission is interrupted, locate the breakpoint through the hash identifier difference and trigger the resume transmission mechanism. Before resuming the transmission, verify the partial write integrity of the remote data block. S4. Receive backup strategy weight parameters sent by the mobile terminal through an encrypted communication channel, and update the entropy weight allocation and dynamic baseline adjustment logic in the multi-source collaborative decision-making unit in real time.

[0008] Compared with the prior art, the beneficial effects of the present invention are as follows: By collecting data on single viewing duration and time period distribution through the user behavior profiling modeling unit, and inputting it into the pre-trained long short-term memory network to generate dynamic profile parameters, the multi-source collaborative decision-making unit calculates the content access probability and dynamically adjusts the threshold to mark high-priority backup objects, achieving the technical effect of accurately screening backup content. This solves the problem of resource waste or omission caused by the rigidity of existing backup strategies, and has the advantages of saving remote storage and network bandwidth resources and avoiding the omission of user-preferred content.

[0009] By combining storage cost, network overhead, and equipment reliability into a multi-dimensional decision matrix through multi-source collaborative decision-making units, information entropy weights are calculated using the weighted entropy method to drive the Dijkstra algorithm to generate the optimal path. When bandwidth fluctuates, a path redundancy mechanism is activated to switch to a backup path, achieving the technical effect of dynamically matching the optimal backup path. This solves the problem of single-path consistent backup failure or delay, and has the advantages of balancing backup cost and reliability and ensuring backup continuity.

[0010] By generating unique hash identifiers from high-priority objects in blocks, the breakpoint is located by comparing the hashes of the local and remote metadata databases when transmission is interrupted. After the network is restored, the transmission is resumed based on the breakpoint and the data integrity is verified. This achieves the technical effect of efficiently handling transmission interruptions, solving the problems of bandwidth waste and long time consumption caused by full retransmission after interruption. It has the advantages of reducing redundant transmission, shortening backup time, and ensuring data integrity. Attached Figure Description

[0011] Figure 1 This is an overall block diagram of the present invention; Figure 2 This is the overall flowchart of the present invention.

[0012] The meanings of the labels in the diagram are as follows: 1. User behavior profiling modeling unit; 2. Multi-source collaborative decision-making unit; 3. Remote configuration interface unit. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] This invention provides a smart TV system that supports remote storage management. Please refer to [link / reference]. Figure 1 As shown, the system includes a user behavior profiling modeling unit 1, which collects user viewing behavior data in real time and inputs it into a pre-trained long short-term memory network to generate user behavior profile parameters. The user behavior profiling modeling unit 1 uses a built-in time-series data collector to capture four types of behavioral data in real time: single viewing duration, daily viewing time distribution, content type switching frequency, and continuous episode skipping interval. The behavioral data is input into the pre-trained long short-term memory network according to a time window sequence. The hidden layer of this network extracts time-segmentation features through a sliding time window mechanism. The output layer fuses the time-segmentation features with the user's historical profile baseline to generate dynamically updated user behavior profile parameters, where the historical profile baseline is stored in a local encrypted cache. The multi-source collaborative decision-making unit 2 performs the following operations based on the user behavior profile parameters, remaining local storage space, real-time home network bandwidth, and remote storage device response latency: (a) Calculate the content access probability value based on user behavior profile parameters. In operation (a), the multi-source collaborative decision-making unit 2 calculates the content access probability value based on the content type preference weight, daily viewing time distribution density, and urgency of following the series in the user behavior profile parameters through the behavior weight factor allocation module. The behavior weight factor allocation module performs a convolution operation on the content type preference weight and the daily viewing time distribution density, and then adds a non-linear correction amount of the urgency of following the series to generate the content access probability value. The preset threshold of the content access probability value adopts a dynamic baseline adjustment mechanism. This mechanism automatically adjusts the threshold range according to the ratio of the remaining space of local storage to the response latency of remote storage devices. When the content access probability value is higher than the upper limit of the dynamically adjusted threshold, it is marked as a high-priority backup object, and at the same time, the probability value decay counter is triggered to reduce the frequency of repeated backup of the same content.

[0015] It needs further explanation that, in order to achieve accurate backup management of user-preferred content in a smart TV system, a dynamic user behavior profile must first be constructed through the user behavior profile modeling unit 1 to provide a core basis for subsequent backup decisions. The specific implementation method is as follows: User behavior profiling modeling unit 1 captures four types of key behavioral data in real time through a built-in time-series data collector. This time-series data collector is a module specifically designed to record changes in user behavior over time, featuring millisecond-level timing accuracy and encrypted data transmission to prevent data loss or leakage. Specifically, the single viewing duration is calculated by monitoring the smart TV's playback status, starting from when the user clicks the play button and ending when they click stop or switch content. If the pause exceeds 5 minutes, it is considered a temporary absence, and the time before the pause is included in the single viewing duration. For example, if the user clicks play at 19:00, the 19:00 viewing duration is calculated as follows: A pause exceeding 5 minutes is recorded as a single viewing session duration of 45 minutes. Daily viewing time distribution is calculated by dividing a 24-hour day into 12 two-hour periods and recording the total viewing time within each period. For example, a user might watch 90 minutes between 7 PM and 9 PM, 30 minutes between 9 PM and 11 PM, and zero minutes in other periods, reflecting their viewing preferences. Content type switching frequency is calculated by recording the number of times a user switches content types within one hour, such as from TV dramas to variety shows or movies. If a user switches 3 times within an hour, the frequency is 3 times / hour; a lower frequency indicates a lower frequency. The higher the user's focus on the current content type, the stronger the user's willingness to follow the series. The interval between episode skips is calculated based on the user's behavior while watching a series. For example, if a user clicks on episode 2 10 seconds after watching episode 1, the interval is recorded as 10 seconds. A shorter interval indicates a stronger willingness to follow the series. After capturing four types of behavioral data, the data is input into a pre-trained Long Short-Term Memory (LSTM) network according to a time window sequence. The time window sequence is formed by dividing the behavioral data from seven consecutive days into 24 one-hour time windows per day, creating 168 (7×24) time window data blocks. Each data block contains data for that time window. The four types of behavioral data, such as the single viewing duration, time period distribution, switching frequency, and jump interval of 8-9 am on October 1st, are arranged into a sequence in chronological order to ensure the temporal correlation of the data. The pre-trained Long Short-Term Memory (LSTM) network is a model pre-trained with 100,000 sets of user behavior data, which can effectively handle long-term dependencies in time-series data and avoid traditional neural networks forgetting early behavioral information. When inputting the sequence, the network first normalizes the data, mapping each type of data to the range of 0-1. For example, the maximum single viewing duration is 180 minutes, and 45 minutes is normalized to 0.25. The data is then fed into the hidden layer. This hidden layer extracts time-segment association features using a sliding time window mechanism. This mechanism involves setting a sliding window containing three consecutive time windows, which slides along the sequence with a step size of one time window. Each time the window slides, the association degree of the three types of behavioral data within the window is calculated. For example, a negative correlation is calculated between the frequency of content type switching and the duration of a single viewing session (low switching frequency means longer viewing time, higher association), and a positive correlation is calculated between the interval between consecutive episode jumps and the time-segment distribution (shorter jump intervals during frequently viewed time periods, higher association). These association degrees are integrated into time-segment association features. If the average association degree within a certain sliding window is 0.8, it indicates strong association and stable preferences for user behavior within that time period. Subsequently, the output layer fuses the time-segment association features with the user's historical profile baseline to generate dynamically updated user behavior profile parameters. The user historical profile baseline is a stable parameter of the user's behavior profile over the past 30 days, containing long-term content type preferences and basic information about fixed viewing time periods. For example, the user's preference weight for TV dramas in the baseline is 0. 7. The fixed viewing period is 19:00-21:00. This baseline is stored in a local encrypted cache. The local encrypted cache is an encrypted storage area built into the smart TV. It uses the AES-256 encryption algorithm to protect user profile data and prevent third parties from stealing or tampering with it. When the output layer is fused, the time period association features are first weighted and fused with the historical profile baseline. For example, the variety show preference weight in the baseline is 0.3, and the variety show association degree in the time period association features is 0.6. After fusion, the variety show preference weight becomes 0.3×0.6+0.6×0.4=0.42. Finally, user behavior profile parameters containing content type preference weight, daily viewing time period distribution density, and urgency of following continuous dramas are generated to realize dynamic updates of the profile and ensure that it can reflect the latest user behavior habits. After the user behavior profile parameters are generated, the multi-source collaborative decision-making unit 2 needs to perform operation (a) based on these parameters to calculate the content access probability value to determine the high-priority backup objects. This is because only by accurately judging the content that users may access again can the waste of backup resources be avoided. The specific implementation method is as follows: In operation (a), the multi-source collaborative decision-making unit 2 first extracts three core indicators from the user behavior profile parameters: content type preference weight, daily viewing time distribution density, and urgency of following consecutive TV series. The content type preference weight is assigned separately for each content type, ranging from 0 to 1, with a total weight of 1. For example, if a user's viewing percentage for TV series in the past 7 days is 60%, variety shows 30%, and movies 10%, then the corresponding preference weights are 0.6, 0.3, and 0.1, respectively. The daily viewing time distribution density is calculated using a 1-hour time window, calculating the proportion of user viewing time within each time window to the total daily viewing time. For example, if a user's total daily viewing time is 120 minutes, the proportion of viewing time within each time window to the total daily viewing time is calculated. Watching 48 minutes at 8 PM results in a density of 48 / 120 = 0.4. The urgency level for following a series is determined by the series' update time and the user's historical follow-up intervals. If the series updates on the same day and the user's past follow-up intervals are no more than 24 hours, the urgency level is set to 0.9. If the series updates more than 3 days ago and the user hasn't watched it, the urgency level is set to 0.3, with a value range of 0-1. The content access probability value is calculated through a behavioral weighting factor allocation module. This module is specifically designed to integrate multi-dimensional behavioral indicators and generate probability values. Its core logic is to first integrate basic preferences and time period characteristics, and then adjust using the follow-up urgency level to ensure that the probability value closely matches the user's actual access intentions. The module first performs a convolution operation on the content type preference weights and the daily viewing time distribution density. This convolution operation is used to mine the correlation features between the two types of indicators. For example, if a user's preference weight for TV dramas is 0.6, and the viewing density between 7 PM and 8 PM is 0.4, the convolution operation will highlight the feature value of the user's access to TV dramas between 7 PM and 8 PM, making the fused features more reflective of the user's tendency to access specific content during specific time periods, avoiding feature fragmentation caused by simple addition. After the convolution operation, a basic correlation feature value is obtained; for example, in the above case, the basic correlation feature value is 0.24. This is then superimposed with a non-linear correction factor for the urgency of following consecutive episodes to generate the content access probability. The non-linear correction value is an adjustment value set according to the non-linear variation of the urgency of following updates. Unlike the fixed proportional adjustment of linear correction, it dynamically changes the correction range according to the urgency range. When the urgency is between 0 and 0.5, the correction value is urgency × 0.2 (to avoid excessive influence from low urgency). When the urgency is between 0.5 and 1, the correction value is urgency × 0.5 (high urgency requires special consideration). For example, if the urgency of following an episode of a certain series is 0.9, the non-linear correction value is 0.9 × 0.5 = 0.45. This correction value is superimposed on the basic correlation feature value of 0.24, and the final content access probability value is 0.24 + 0.45 = 0.69. The closer this value is to 1, the higher the probability that a user will access the content in the future. After obtaining the content access probability value, it is necessary to determine whether to mark it as a high-priority backup object through a preset threshold. However, a fixed threshold may not be adaptable due to changes in the remaining local storage space and the response latency of remote storage devices. For example, when local storage is sufficient and remote response is fast, a fixed threshold that is too high may miss some content that users may access. When local storage is scarce and remote response is slow, a threshold that is too low may result in backing up too much invalid content. Therefore, a dynamic baseline adjustment mechanism is required. The specific implementation method is as follows: The preset threshold for content access probability adopts a dynamic baseline adjustment mechanism. This mechanism is a strategy that adjusts the threshold range in real time based on the local storage status of the smart TV and the performance of the remote storage device. Its core is to quantify the current compatibility between storage and remote devices by the ratio of remaining local storage space to the response latency of the remote storage device, avoiding the bias of a single-dimensional judgment. The remaining local storage space is obtained in real time through the smart TV's storage management module, and the remote storage device response latency is obtained by sending test data packets to the remote device and recording the round-trip time. The ratio is calculated as: remaining local storage space ÷ remote storage device response latency. A higher ratio indicates more sufficient local storage and faster remote response, allowing the threshold range to be appropriately lowered, enabling more potentially high-access-probability content to enter the backup scope. A lower ratio requires raising the threshold range to strictly filter high-priority content. The specific rules for automatically adjusting the threshold range are as follows: When the ratio is ≥0.3GB / ms (sufficient local storage, fast remote access), the threshold range is set to 0.5-0.7, meaning a probability value higher than 0.7 is marked as high priority. When the ratio is between 0.1-0.3GB / ms (in progress), the threshold range is set to 0.6-0.8. When the ratio is <0.1GB / ms (limited local storage, slow remote access), the threshold range is set to 0.7-0.9. For example, if at a certain moment local storage has 8GB remaining and remote response latency is 100ms, the ratio is 0.08GB / ms, and the threshold range is adjusted to 0.7-0.9. If the access probability of a certain content is 0.85, which is lower than 0.9, it is not marked. If the probability value is 0.92, which is higher than 0.9, it is marked as a high-priority backup object. Simultaneously, to avoid repeated backups of the same content due to a probability value consistently exceeding the threshold (e.g., if a user has already backed up a certain episode and the subsequent probability value remains high, causing repeated backups and consuming resources), a probability decay counter needs to be triggered. To reduce the frequency of duplicate backups of the same content, a probability decay counter is a counting module activated for content already marked as high priority. The counter starts at 0 and increments by 1 every 24 hours, with each count corresponding to a probability decay of 0.05. The decay rate has been experimentally calibrated to avoid rapid decay leading to missed detections and effectively reduce duplicate backups. For example, when a series is first marked as high priority, its probability value is 0.92 and the counter is 0. After 24 hours, the counter is 1, and the probability value decays to 0.87. If the upper threshold is still 0.9, the probability value of 0.87 is lower than the upper threshold and is no longer marked as high priority, thus avoiding duplicate backups. If the counter accumulates to 5 times (after 120 hours), the probability value decays to 0.92 - 5 × 0.05 = 0.67, further reducing the risk of duplicate backups. At the same time, if a user accesses the content again later, the behavior profile parameters will be updated, and the probability value will increase again, ensuring that content with genuine access needs can still be marked.

[0016] Content with access probability values ​​higher than a preset threshold is marked as high-priority backup objects. (b) A multi-dimensional decision matrix including storage cost, network overhead, and device reliability is constructed based on the remaining local storage space, real-time home network bandwidth, and remote storage device response latency. When constructing the multi-dimensional decision matrix in operation (b), the storage cost is quantified by the product of the remaining local storage space occupancy rate and the unit capacity cost of the remote storage device. The network overhead is calculated based on the weighted sum of the home network real-time bandwidth fluctuation coefficient and the number of hops in the transmission path. The device reliability is generated based on the stability index of the remote storage device response latency and the historical failure rate. The three sets of dimensional data are standardized to form a multi-dimensional decision matrix. The multi-dimensional decision matrix is ​​combined with the weighted entropy method to dynamically generate the optimal backup path. The multi-dimensional decision matrix and the weighted entropy method are then combined. When dynamically generating the optimal backup path, the algorithm calculates the information entropy weights of three dimensions: storage cost, network overhead, and device reliability using the entropy method. Then, it weights and aggregates the decision matrix based on these entropy weights to generate path decision values. Finally, the path decision values ​​are imported into the Dijkstra optimization algorithm to output the optimal backup path connecting local storage and the target remote storage device. During the path decision value aggregation process, if the Dijkstra optimization algorithm detects fluctuations in the real-time bandwidth of the home network exceeding a preset tolerance range, it automatically activates a path redundancy assessment mechanism. This mechanism includes: generating backup path decision values ​​based on historical network bandwidth data; comparing the difference coefficient between the primary and backup path decision values ​​with the device reliability score; and switching to the backup path when the difference coefficient exceeds the reliability score, with the switching event fed back to the entropy method weight update module.

[0017] It needs further explanation that after marking high-priority backup objects, the multi-source collaborative decision-making unit 2 needs to determine the optimal backup path through operation (b). The selection of this path needs to comprehensively consider the cost and reliability of storage, network, and equipment to avoid low backup efficiency or resource waste caused by a single-dimensional decision. Therefore, a multi-dimensional decision matrix needs to be constructed first. The specific implementation method is as follows: When constructing the multi-dimensional decision matrix in operation (b), the multi-source collaborative decision-making unit 2 first quantifies the storage cost. This cost needs to be calculated by combining local storage pressure and remote storage economics, and is obtained by multiplying the local storage remaining space occupancy rate by the unit capacity cost of the remote storage device. The local storage remaining space occupancy rate is an indicator reflecting the tightness of local storage, which is calculated in real time by the smart TV storage management module. It is calculated as (total local storage capacity - local storage remaining space) ÷ total local storage capacity × 100%. For example, if the total local storage capacity is 100GB and the remaining space is 15GB, then the occupancy rate is (100-15) ÷ 100 × 100%. =85%, the unit capacity cost of remote storage devices is the fee charged by remote storage service providers per unit capacity, which is obtained by remote configuration interface unit 3 from the metadata of remote storage devices. For example, if the unit capacity cost of a certain remote device is 0.5 yuan per GB per month, the product of the two is the storage cost, such as 85% × 0.5 yuan / GB = 0.425 yuan / GB. The higher this value, the more strained the local storage is, the higher the cost of remote storage, and the lower the decision priority of the storage dimension. Next, the network overhead is calculated. The network overhead needs to reflect the bandwidth stability and transmission path complexity, and is determined by the weighted sum of the real-time bandwidth fluctuation coefficient of the home network and the number of hops in the transmission path.The real-time bandwidth fluctuation coefficient of a home network is an indicator of bandwidth stability. It is calculated by collecting real-time bandwidth data over one minute using the network monitoring module of the smart TV, and then calculating (maximum bandwidth - minimum bandwidth) ÷ average bandwidth. For example, if the maximum bandwidth is 120Mbps, the minimum is 80Mbps, and the average is 100Mbps, the fluctuation coefficient is (120-80) ÷ 100 = 0.4. A higher coefficient indicates more unstable bandwidth and a higher transmission risk. The transmission path hop count refers to the number of network routing nodes from the smart TV to the remote storage device, obtained using the traceroute command. For example, if the path from the TV to the remote device involves 5 nodes (TV - home router - community gateway - operator node - remote storage gateway), the hop count is 5. A higher hop count indicates higher data transmission latency and packet loss probability. In the weighted sum calculation, the network impact weights are preset (bandwidth fluctuation has a greater impact on transmission, so the weight is set to 0.6; the hop count weight is set to 0.4). For example, with a fluctuation coefficient of 0.4 and a hop count of 5, the network overhead is 0.4 × 0.6. +5×0.4=0.24+2=2.24. The higher this value, the higher the transmission cost and risk at the network level. Then, the device reliability is generated. Device reliability directly determines the security of backup data. A reliability score is generated based on the stability index of the remote storage device's response latency and the historical failure rate. The stability index of the remote storage device's response latency is an indicator to evaluate the consistency of the remote device's response. Response latency data of the remote device is collected within 30 minutes (collected once every minute, for a total of 30 data points). The standard deviation of latency is divided by the average latency. For example, if the average latency is 60ms and the standard deviation is 6ms, then the stability index is 6÷60=0.1. The smaller the index, the more stable the response. The historical failure rate is the failure frequency of the remote device in the past 30 days, which is statistically analyzed by the remote storage device management platform. That is, the total failure time ÷ the total runtime × 100%. For example, if there were 2 failures in the past 30 days, with a cumulative failure time of 1 hour and a total runtime of 720 hours, then the historical failure rate is 1÷720×100%≈0.14%. The lower the failure rate, the more reliable the device.When calculating the reliability score, the stability index and historical failure rate are first mapped to the range of 0-1 (index 0.1→0.1, failure rate 0.14%→0.14). Then, the score is calculated as 1 - (stability index × 0.4 + historical failure rate × 0.6) (stability has a greater impact on real-time backup, with a weight of 0.4; failure rate has a greater impact on long-term storage, with a weight of 0.6). For example, 0.1 × 0.4 + 0.14 × 0.6 = 0.04 + 0.084 = 0.124, and the reliability score is 1 - 0.124 = 0.876. The closer the score is to 1, the higher the equipment reliability. The three sets of data were standardized. Standardization maps storage costs, network overhead, and device reliability data with different dimensions and ranges to a uniform 0-1 range. This prevents the impact of any one dimension on decision-making from being excessively amplified due to differences in data range. Specifically, storage costs and network overhead are mapped using (original value - minimum value) ÷ (maximum value - minimum value) (the larger the original value, the larger the standardized value, representing higher costs). Device reliability is mapped using (original value - minimum value) ÷ (maximum value - minimum value) (the larger the original value, the larger the standardized value, representing higher reliability). (The higher the value), for example, the original value of storage cost is 0.425 yuan / GB (minimum 0.1, maximum 0.8), after standardization it is (0.425-0.1)÷(0.8-0.1)≈0.46; the original value of network overhead is 2.24 (minimum 1, maximum 5), after standardization it is (2.24-1)÷(5-1)≈0.31; the original value of device reliability is 0.876 (minimum 0.5, maximum 1), after standardization it is (0.876-0.5)÷(1-0.5)≈0.75. These three sets of standardized data are then divided into storage cost - network overhead - device reliability. The reliability of backup devices is arranged in order, forming a vector such as [0.46, 0.31, 0.75]. If there are multiple candidate remote storage devices, each device corresponds to a vector. The combination of all vectors forms a multidimensional decision matrix, which provides basic data for subsequent optimal path calculation. After the multidimensional decision matrix is ​​constructed, each dimension needs to be assigned a reasonable weight using the weighted entropy method, and then combined with the optimization algorithm to generate the optimal backup path. This is because the importance of each dimension in the matrix to path selection is different, and the weights need to be objectively calculated using the entropy method to avoid decision bias caused by subjective weight setting. The specific implementation method is as follows: When dynamically generating optimal backup paths by combining a multidimensional decision matrix with the weighted entropy method, the information entropy weights for three dimensions—storage cost, network overhead, and device reliability—are first calculated using the entropy method. The entropy method is an objective method for determining weights based on the degree of data dispersion. Higher data dispersion, meaning greater differences between different candidate paths in that dimension, indicates richer decision information and thus a higher weight. The information entropy weights are the weight values ​​for each dimension calculated using the entropy method, with a total weight of 1. Specifically, the information entropy is first calculated for each dimension of the multidimensional decision matrix, such as the standardized storage cost values ​​of all candidate paths. For example, if there are three candidate path data points for a certain dimension, namely 0.46, 0.52, and 0.61, first calculate the proportion of each data point to the total data in that dimension (e.g., 0.46 ÷ (0.46 + 0.52 + 0.61) ≈ 0.32), then substitute them into the entropy formula to calculate the information entropy of that dimension. Next, calculate the difference coefficient based on the information entropy (difference coefficient = 1 - information entropy). The larger the difference coefficient, the higher the weight. Finally, normalize the difference coefficients of each dimension to obtain the information entropy weight. For example, the difference coefficients for storage cost dimension are 0.35, network overhead is 0.42, and device reliability is 0.23. After normalization, the weights are 0.35 ÷ (0.46 + 0.52 + 0.61) ≈ 0.32. The formula (35 + 0.42 + 0.23) = 0.35, 0.42, 0.23 indicates that network overhead has the greatest impact on path decision-making, while device reliability has the least impact. Next, the decision matrix is ​​weighted and aggregated based on information entropy weights to generate path decision values. Weighted aggregation multiplies the standardized dimensional data of each candidate path by its corresponding information entropy weight and then sums the results to obtain the comprehensive decision value for that path. The magnitude of the decision value needs to be determined in conjunction with the dimensional attributes. Storage cost and network overhead are cost-related indicators (lower values ​​are better), while device reliability is a benefit-related indicator (higher values ​​are better). Therefore, during aggregation, it is necessary to ensure that the decision value comprehensively reflects low cost and high reliability. For a high optimal state, for example, if the standardized data of a candidate path is [0.46 (storage cost), 0.31 (network overhead), 0.75 (device reliability)], and the information entropy weight is [0.35, 0.42, 0.23], then the path decision value is (1-0.46)×0.35+(1-0.31)×0.42+0.75×0.23, where (1-cost data) is the conversion of cost indicators into benefit indicators (the higher the value, the better). The calculated value is 0.54×0.35+0.69×0.42+0.75×0.23≈0.189+0.2898+0.1725≈0.6513. The closer the decision value is to 1, the more optimal the overall path is. Finally, the path decision value is imported into Dijkstra's algorithm to output the optimal backup path connecting local storage and the target remote storage device. Dijkstra's algorithm is a classic shortest path algorithm, used here to find the path with the largest decision value (overall optimal) among the candidate paths. The algorithm first uses the decision values ​​of all candidate paths as path weights, local storage as the starting point, and each candidate remote storage device as the ending point to construct a path graph. Then, starting from the starting point, it calculates the path weight from the starting point to each ending point in turn, i.e., the path decision value, and records the current optimal path. When all... After the path weights of the destination are calculated, the path with the highest weight is selected as the optimal backup path. The algorithm will output the specific transmission nodes of this path, such as local storage → home router → carrier backbone node → remote storage device A, to ensure that backup data can be transmitted along the optimal path, balancing cost and reliability. During the path decision value aggregation process of Dijkstra's optimization algorithm, drastic fluctuations in the real-time bandwidth of the home network may cause the selected primary path to suddenly become unreliable, such as a sudden drop in bandwidth causing a surge in transmission latency. If not adjusted in time, it will cause backup interruption or data loss. Therefore, it is necessary to automatically enable the path redundancy evaluation mechanism. The specific implementation method is as follows: During the path decision aggregation process, Dijkstra's optimization algorithm first monitors the real-time bandwidth fluctuation of the home network. Fluctuation range refers to the percentage of the difference between the maximum and minimum real-time bandwidth within 10 seconds relative to the average bandwidth. The preset tolerance range is the upper limit of bandwidth fluctuation allowed by the system, experimentally calibrated to be 20%. That is, when the fluctuation range exceeds 20%, the bandwidth is considered unstable. For example, if the average bandwidth is 100Mbps, and the maximum value within 10 seconds is 110Mbps and the minimum is 75Mbps, the fluctuation range is (110-75)÷100×100%=35%, exceeding the preset tolerance range of 20%. At this point, the algorithm automatically activates the path redundancy assessment mechanism, which specifically includes: The first step is to generate alternative path decision values ​​based on historical network bandwidth data. Historical network bandwidth data refers to the bandwidth fluctuation coefficient and transmission path hop count of the same time period cached by the smart TV over the past 7 days. For example, if the current time is 20:00, the data is cached for the bandwidth fluctuation coefficient and transmission path hop count from 20:00 to 20:10 over the past 7 days. Combined with the storage cost and device reliability data of the alternative path in the current multi-dimensional decision matrix, and using the weighted sum and reliability scoring method described earlier, the path decision values ​​for the alternative path are regenerated. For example, if the alternative path corresponds to remote storage device B, with a historical bandwidth fluctuation coefficient of 0.3, a hop count of 4, a standardized storage cost value of 0.51, and a device reliability of 0.82, then... With the information entropy weight unchanged, the calculated backup path decision value is approximately 0.682. The second step compares the difference coefficient between the primary and backup path decision values ​​with the device reliability score. The decision value difference coefficient is an indicator that measures the difference between the primary and backup path decision values. It is calculated as |primary path decision value - backup path decision value| ÷ primary path decision value × 100%. For example, if the primary path decision value is 0.715 and the backup path decision value is 0.682, the difference coefficient is |0.715 - 0.682| ÷ 0.715 × 100% ≈ 4.6%. The device reliability score is the reliability score of the remote storage device corresponding to the current primary path (e.g., 0.876). During comparison, if the difference coefficient is less than or equal to the equipment reliability score (e.g., 4.6% ≤ 87.6%), it indicates that the difference between the primary and backup paths is small, and the primary equipment has high reliability, so no switching is needed. If the difference coefficient is greater than the equipment reliability score, such as the decision value of the primary path dropping to 0.62 due to bandwidth fluctuations, and the backup path to 0.682, the difference coefficient is approximately 10%, exceeding the equipment reliability score of 8.76%. In this case, it is determined that the primary path can no longer meet the optimal transmission requirements, and a switch to the backup path is necessary. The third step is to switch to the backup path when the difference coefficient exceeds the reliability score, and the switching event is fed back to the entropy method weight update module. During the switch, the multi-source collaborative decision-making unit 2 immediately sends a data... The transmission module sends a path switching command, stopping primary path data transmission and starting backup path transmission, such as switching from local → device A to local → device B. Simultaneously, it records the switching time, primary / backup path information, and the switching reason log. The entropy weight update module, responsible for dynamically adjusting the information entropy weights, records the path switching caused by this bandwidth fluctuation as a signal of increased network overhead. In the next calculation of information entropy weights, it appropriately increases the weight of the network overhead dimension to ensure that subsequent path decisions prioritize network stability, reduce frequent switching due to bandwidth fluctuations, and achieve adaptive optimization of path decisions.

[0018] (c) When the real-time bandwidth of the home network is detected to be higher than the preset network bandwidth for a continuous fixed period of time and the remote storage device is idle, a block-based incremental backup of the high-priority backup object is triggered. The execution process of the block-based incremental backup in operation (c) is as follows: High-priority backup objects are divided into fixed-size data blocks according to content type. A unique hash identifier is generated for each data block. When the remote storage device is detected to be idle, only incremental data blocks not recorded in the local hash identifier library are transmitted. After the transmission is completed, the new hash identifier is synchronized to the metadata database of the remote storage device. If a transmission interruption occurs during the block incremental backup process, the remote storage management module locates the data blocks that have not been transmitted and records the breakpoint by comparing the hash identifier differences between the local and remote metadata databases. When the network bandwidth recovers to a stable state, the transmission resume mechanism is triggered based on the breakpoint location. Before resuming the transmission, the partial write integrity of the target data block in the remote storage device is verified first.

[0019] Further explanation is needed: after determining the optimal backup path connecting the local and remote storage devices using Dijkstra's optimization algorithm, the multi-source collaborative decision-making unit 2 initiates the block-based incremental backup process in operation (c). Compared to traditional full backups, which require repeated transmission of backed-up data, block-based incremental backups only transmit newly added data blocks that have not yet been backed up. This significantly reduces network bandwidth usage and backup time while ensuring the integrity of data transmission. The specific implementation method is as follows: The execution process of block incremental backup in operation (c) is as follows: First, high-priority backup objects are divided into fixed-size data blocks based on content type. Content type here encompasses the specific data categories of the high-priority backup objects, such as TV dramas and variety show videos, and user-collected TXT and PDF documents. Different data types have significantly different structures and compression characteristics. Video data is large and highly continuous, while document data is small and fragmented. Segmenting by content type adapts to different data characteristics to reduce transmission errors. The fixed size is dynamically set according to the content type: each data block for video content is fixed at 100MB to balance the stability of a single transmission with the frequency of requests; each data block for document content is fixed at 10MB to avoid the overhead of frequent requests caused by small-block transmissions. The segmentation operation is performed by the smart TV. The built-in file splitting module executes the process. This module reads the binary data of high-priority backup objects segment by segment according to a preset number of bytes. For example, an 800MB TV series file is split into 100MB segments, generating 8 data blocks. Each block is appended with a unique block number and its corresponding file identifier to ensure accurate correspondence with the original file during subsequent data reassembly and to avoid block order confusion. Next, a unique hash identifier is generated for each data block. This hash identifier is a 64-bit string calculated using the SHA-256 hash algorithm on the complete binary content of the data block. Its core characteristic is that different data blocks correspond to different hash identifiers, and identical data blocks correspond to unique hash identifiers. This allows for precise determination of whether data blocks are duplicated or tampered with. During the generation process, the file splitting module calls the function after each data block is output. The local hash calculation unit performs calculations on the binary content of the block, and simultaneously associates and stores the hash identifier with the block sequence number, the file name, and the generation time of the data block, forming a block-hash mapping relationship. This provides a basis for subsequent incremental judgments. Then, when the remote storage device is detected to be idle (meaning the core resource utilization rate of the remote storage device is lower than a preset threshold, CPU utilization ≤ 30%, disk I / O load ≤ 20%, network I / O load ≤ 25%), the multi-source collaborative decision-making unit 2 obtains the real-time status data of the device by sending periodic heartbeat packets (every 5 seconds) to the remote storage device. If the resource utilization rate returned by three consecutive heartbeat packets is lower than the threshold, the device is determined to be in an idle state. If the remote device is in a busy state... If the CPU utilization reaches 50%, the system will re-check every 10 seconds until the device is idle. This avoids packet loss or increased latency caused by transmitting data when the device is under high load. Subsequently, only incremental data blocks not recorded in the local hash identifier database will be transmitted. The local hash identifier database is an SQLite database stored locally on the smart TV. It specifically records the hash identifiers of all data blocks that have been successfully backed up to the remote storage device. The database table structure includes fields for the file backup time and block sequence number to which the hash identifier belongs, ensuring that information on backed-up blocks can be quickly retrieved. Before transmission, the multi-source collaborative decision-making unit 2 first compares all hash identifiers of the data blocks to be backed up with the records in the local hash identifier database one by one. If the hash identifier of a data block has no matching record in the database, it means that the block has not been backed up.These are then marked as incremental data blocks. For example, if there are 8 data blocks to be backed up, and 6 of them already exist in the local database while 2 do not, only these 2 incremental data blocks will be transmitted. This effectively reduces redundant transmission and saves network bandwidth. Finally, after the transmission is complete, the new hash identifiers are synchronized to the metadata database of the remote storage device. The metadata database of the remote storage device is a structured MySQL database used by the remote storage end to manage all backup data blocks. It records the hash identifier, storage path, associated high-priority backup object name, and reception time information of the data blocks, ensuring that the data blocks can be quickly located during subsequent data reading and verification. During synchronization, the smart TV uses TLS 1.3 encryption. A secure HTTP POST request sends the hash identifier, file name, and block sequence number of the newly transmitted data block to the remote metadata database. Upon receiving the request, the remote metadata database first queries the storage path of the corresponding data block to confirm the file's complete existence. Then, it writes the new hash identifier and associated information into the database table, completing synchronization and ensuring consistency between local and remote hash records. This provides an accurate reference for the next incremental backup. During block-based incremental backup, fluctuations in the home network may cause transmission interruptions. Directly retransmitting all data blocks would waste resources. Therefore, a targeted resumption mechanism needs to be implemented through the remote storage management module to ensure complete data backup. The specific implementation method is as follows: If a transmission interruption occurs during incremental backup, the smart TV first determines the interruption status. After sending a data block to the remote storage device, if the smart TV does not receive a successful reception confirmation packet within 30 seconds, or receives a reception failure packet (such as a data verification error), it immediately stops the current data block transmission and records key information at the time of the interruption, including the hash identifier of the data block being transmitted and the number of bytes transmitted. Subsequently, the remote storage management module compares the hash identifiers of the local and remote metadata databases. The remote storage management module is a built-in functional unit of the smart TV, specifically responsible for interacting with the remote storage device, including data transmission control, device status monitoring, and data verification. During the comparison, this module first sends a hash list retrieval request to the remote storage device, which is then received by the remote metadata database. The module then returns a list of hash identifiers for all data blocks currently stored and related to the backup target, including those marked as fully received and those not fully received. Next, the module compares the hash list returned remotely with the local hash list of the data blocks to be backed up, filtering out two types of discrepancies: first, hash identifiers present locally but not recorded in the remote metadata database, corresponding to incremental blocks that have not yet started transmission; second, hash identifiers present both locally and in the remote metadata database but marked as not fully received, corresponding to incomplete blocks whose transmission was interrupted. The module then locates the incomplete data blocks and records the breakpoint. When locating incomplete blocks, the remote storage management module uses the hash identifiers marked as not fully received in the remote metadata database to find the corresponding incomplete data block. The breakpoint location refers to the location of the incomplete data block. The remote metadata database records the number of bytes successfully transmitted to the remote storage device for each incompletely received data block. The remote storage management module retrieves this byte count via a breakpoint query request and marks a breakpoint at the corresponding location of the local data block to be backed up. For example, in the local data block's binary file, it marks the start point as pending resumption of transmission from byte 65MB+1, ensuring that subsequent resumptions only target the unreceived portion. Then, when the network bandwidth returns to a stable state (determined by the smart TV's network monitoring module), the module collects real-time bandwidth data every 5 seconds and calculates the fluctuation range of the difference between the maximum and minimum bandwidth values ​​within 10 seconds relative to the average bandwidth. If the fluctuation range is collected for 6 consecutive times (30 seconds in total), the calculation is performed accordingly. If the fluctuation is ≤10%, the network is considered to have stabilized. If the fluctuation still exceeds 10%, monitoring continues to prevent restarting the transmission during network instability, which could lead to another interruption. Subsequently, a transmission resumption mechanism is triggered based on the breakpoint location. When the transmission resumption mechanism is activated, the remote storage management module reads binary data from the breakpoint of the local data block to be backed up (e.g., 65MB + 1 byte) according to the breakpoint location, generating a transmission resumption data segment (e.g., for a 100MB data block, the transmission resumption data segment is from 65MB + 1 byte to 100MB, totaling 35MB). At the same time, a transmission resumption request is sent to the remote storage device. The request includes the hash identifier of the incomplete data block, the breakpoint location, and the byte range of the transmission resumption data segment. After receiving the request, the remote storage device will prepare to receive the transmission resumption data at the breakpoint of the corresponding data block.To ensure data block continuity and avoid duplicate transmission of received portions, the final step before retransmission prioritizes verifying the partial write integrity of the target data block on the remote storage device. Partial write integrity refers to verifying whether the content of the received portion of the data block on the remote storage device matches the corresponding portion of the local data block to be backed up, preventing corruption of remotely received data due to transmission interruptions. During verification, the remote storage management module first sends a partial data hash request to the remote storage device to obtain the binary content of the remotely received data block (e.g., the 65MB portion) and calculates its SHA-256 hash value. Simultaneously, the local module reads the corresponding byte range (0 to 65MB) from the data block to be backed up. The binary content of the data block is also processed, and its SHA-256 hash value is calculated. If the hash values ​​calculated in both locations match, the data is considered partially written and can be resumed. If the hash values ​​do not match, the remote storage management module sends a data cleanup request to the remote device, deleting the received portion of the data block from the remote device. The data is then retransmitted from the beginning of the data block to ensure the accuracy of the resumed data and prevent concatenated or damaged data blocks from affecting subsequent use. Finally, it connects to the remote configuration interface unit 3, which provides an encrypted communication channel to the mobile terminal, receives manually adjusted backup strategy weight parameters, and synchronizes them to the multi-source collaborative decision-making unit 2, improving the effectiveness of remote storage management.

[0020] This invention collects user viewing behavior data, generates behavioral profile parameters through a pre-trained long short-term memory network, calculates content access probabilities based on these parameters, marks high-priority backup objects, constructs a multi-dimensional decision matrix by combining storage and bandwidth, generates the optimal backup path using the weighted entropy method and Dijkstra's algorithm, performs block incremental backup, locates breakpoints and resumes transmission after interruption by hash comparison, and updates decision logic by receiving parameters from mobile terminals. This invention solves the problems of wasted backup resources and poor path adaptation, and improves the efficiency and reliability of remote storage management for smart TVs.

[0021] The second objective of this invention is to provide a method for implementing a smart TV system supporting remote storage management, including any of the above-mentioned features, comprising the following steps: S1. Capture single viewing duration, daily viewing time distribution, content type switching frequency, and continuous episode jump interval through a time series data collector. Input the behavioral data into a pre-trained long short-term memory network according to the time window sequence. The hidden layer of the network extracts time-related features and fuses them with the historical profile baseline to dynamically generate user behavior profile parameters. S2. Calculate the content access probability value based on user behavior profile parameters, mark high-priority backup objects in combination with dynamic baseline adjustment mechanism, and construct a multi-dimensional decision matrix based on local storage remaining space, home network real-time bandwidth and remote storage device response latency. Calculate information entropy weights through weighted entropy value method and drive Dijkstra optimization algorithm to generate the optimal backup path. S3. Divide high-priority backup objects into fixed-size data blocks and generate unique hash identifiers. When the real-time bandwidth of the home network is detected to be stable and the remote storage device is idle, transmit incremental data blocks in the local hash identifier library. If the transmission is interrupted, locate the breakpoint through the hash identifier difference and trigger the resume transmission mechanism. Before resuming the transmission, verify the partial write integrity of the remote data block. S4. Receive backup strategy weight parameters sent by the mobile terminal through an encrypted communication channel, and update the entropy weight allocation and dynamic baseline adjustment logic in the multi-source collaborative decision-making unit in real time.

[0022] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart TV system supporting remote storage management, characterized in that, It includes a user behavior profile modeling unit (1), which collects user viewing behavior data in real time and inputs it into a pre-trained long short-term memory network to generate user behavior profile parameters. The multi-source collaborative decision-making unit (2) performs the following operations based on the user behavior profile parameters, the remaining local storage space, the real-time bandwidth of the home network, and the response latency of the remote storage device: (a) Calculate the content access probability value based on user behavior profile parameters, and mark the content access probability value higher than the preset threshold as high priority backup objects; (b) Construct a multi-dimensional decision matrix that includes storage cost, network overhead, and device reliability based on the remaining local storage space, real-time bandwidth of the home network, and response latency of remote storage devices, and dynamically generate the optimal backup path by combining the multi-dimensional decision matrix with the weighted entropy method; (c) When the real-time bandwidth of the home network is detected to be higher than the preset network bandwidth for a fixed period of time and the remote storage device is idle, trigger the block incremental backup of the high-priority backup object; The remote configuration interface unit (3) is used to provide an encrypted communication channel to the mobile terminal, receive the backup strategy weight parameters manually adjusted by the user and synchronize them to the multi-source collaborative decision-making unit (2).

2. The smart TV system supporting remote storage management according to claim 1, characterized in that: The user behavior profile modeling unit (1) captures four types of behavioral data in real time: single viewing duration, daily viewing time distribution, content type switching frequency, and continuous drama series jump interval through the built-in time series data collector. The behavioral data is input into a pre-trained long short-term memory network according to the time window sequence. The hidden layer of the network extracts time period association features through the sliding time window mechanism. The output layer integrates the time period association features with the user's historical profile baseline to generate dynamically updated user behavior profile parameters. The historical profile baseline is stored in a local encrypted cache area.

3. The smart TV system supporting remote storage management according to claim 1, characterized in that, In operation (a), the multi-source collaborative decision-making unit (2) calculates the content access probability value based on the content type preference weight, daily viewing time distribution density and the urgency of following the series in the user behavior profile parameters through the behavior weight factor allocation module. The behavior weight factor allocation module performs a convolution operation on the content type preference weight and the daily viewing time distribution density, and then adds a non-linear correction amount of the urgency of following the series to generate the content access probability value.

4. The smart TV system supporting remote storage management according to claim 3, characterized in that: The preset threshold for the content access probability value adopts a dynamic baseline adjustment mechanism. This mechanism automatically adjusts the threshold range based on the ratio of the remaining local storage space to the response latency of the remote storage device. When the content access probability value is higher than the upper limit of the dynamically adjusted threshold, it is marked as a high-priority backup object, and a probability value decay counter is triggered to reduce the frequency of repeated backups of the same content.

5. The smart TV system supporting remote storage management according to claim 1, characterized in that: When constructing a multidimensional decision matrix in operation (b), the multi-source collaborative decision unit (2) quantifies the storage cost by multiplying the local storage remaining space occupancy rate with the unit capacity cost of the remote storage device. The network overhead is calculated based on the weighted sum of the home network real-time bandwidth fluctuation coefficient and the number of hops in the transmission path. The device reliability is generated based on the stability index of the response delay of the remote storage device and the historical failure rate. The three sets of data are standardized to form a multidimensional decision matrix.

6. The smart TV system supporting remote storage management according to claim 5, characterized in that: When dynamically generating the optimal backup path by combining the multidimensional decision matrix with the weighted entropy method, the information entropy weights of the three dimensions of storage cost, network overhead and device reliability are calculated by the entropy method. Then, the decision matrix is ​​weighted and aggregated according to the information entropy weights to generate the path decision value. Finally, the path decision value is imported into the Dijkstra optimization algorithm to output the optimal backup path connecting the local storage and the target remote storage device.

7. The smart TV system supporting remote storage management according to claim 6, characterized in that: During the path decision value aggregation process, if the fluctuation range of the real-time bandwidth of the home network exceeds the preset tolerance range, the Dijkstra optimization algorithm automatically activates the path redundancy assessment mechanism. Specifically, this includes: generating backup path decision values ​​based on historical network bandwidth data; comparing the difference coefficient between the decision values ​​of the primary path and the backup path with the device reliability score; switching to the backup path when the difference coefficient exceeds the reliability score; and feeding back the switching event to the entropy weight update module.

8. The smart TV system supporting remote storage management according to claim 1, characterized in that: The execution process of block incremental backup in operation (c) is as follows: High-priority backup objects are divided into fixed-size data blocks according to content type. A unique hash identifier is generated for each data block. When the remote storage device is detected to be idle, only incremental data blocks not recorded in the local hash identifier library are transmitted. After the transmission is completed, the new hash identifier is synchronized to the metadata database of the remote storage device.

9. The smart TV system supporting remote storage management according to claim 6, characterized in that: If a transmission interruption occurs during the incremental backup process, the remote storage management module locates the data block that has not been transmitted and records the breakpoint by comparing the hash identifier difference between the local and remote metadata databases. When the network bandwidth recovers to a stable state, the transmission resume mechanism is triggered based on the breakpoint. Before resuming transmission, the partial write integrity of the target data block in the remote storage device is verified first.

10. A control method for implementing a smart television system supporting remote storage management as described in any one of claims 1-9, characterized in that: Includes the following steps: S1. Capture single viewing duration, daily viewing time distribution, content type switching frequency, and continuous episode jump interval through a time series data collector. Input the behavioral data into a pre-trained long short-term memory network according to the time window sequence. The hidden layer of the network extracts time-related features and fuses them with the historical profile baseline to dynamically generate user behavior profile parameters. S2. Calculate the content access probability value based on user behavior profile parameters, mark high-priority backup objects in combination with dynamic baseline adjustment mechanism, and construct a multi-dimensional decision matrix based on local storage remaining space, home network real-time bandwidth and remote storage device response latency. Calculate information entropy weights through weighted entropy value method and drive Dijkstra optimization algorithm to generate the optimal backup path. S3. Divide high-priority backup objects into fixed-size data blocks and generate unique hash identifiers. When the real-time bandwidth of the home network is detected to be stable and the remote storage device is idle, transmit incremental data blocks in the local hash identifier library. If the transmission is interrupted, locate the breakpoint through the hash identifier difference and trigger the resume transmission mechanism. Before resuming the transmission, verify the partial write integrity of the remote data block. S4. Receive backup strategy weight parameters sent by the mobile terminal through an encrypted communication channel, and update the entropy weight allocation and dynamic baseline adjustment logic in the multi-source collaborative decision-making unit in real time.

Images