Network optimization method and device of set top box, electronic equipment and storage medium

Abstract

The invention relates to the technical field of Internet, and provides a network optimization method and device for a set top box, electronic equipment and a storage medium. The method comprises the following steps: sensing and fusing an initial environment data set and set top box operation data to generate environment and equipment joint data, and performing channel frequency band optimization processing according to the environment and equipment joint data to generate a channel resource allocation column; service optimization decision processing is carried out according to the real-time flow sequence of the set top box and the channel resource allocation list to obtain a service optimization strategy, and heat dissipation and power consumption cooperative control is carried out according to a temperature threshold value and temperature change data collected in real time to generate a heat management instruction set; and performing network resource joint scheduling according to the service optimization strategy and the thermal management instruction set to generate a global optimization strategy. Through multi-source information fusion, thermal management control and a cross-domain scheduling mechanism, the network performance and response speed of the set top box are improved, and energy consumption control and stability guarantee of an equipment level are enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet, in particular to a network optimization method and device of a set-top box, an electronic device and a storage medium. BACKGROUND

[0002] The set-top box often faces the problem of unstable network connection in actual use, and the main causes of the fault are the superposition of various factors. On the one hand, the physical environment interference is relatively complex, for example, the interference signal emitted by the adjacent wireless device, the blockage of the signal propagation by the building structure, the influence of the temperature and humidity change on the hardware stability, etc., which will cause the signal quality to decrease; on the other hand, the running state of the set-top box itself can also cause the fault, such as high load running of the processor, too high temperature, aging of the network module, poor heat dissipation, etc., which can cause frequent disconnection, lag or connection failure. In addition, the dynamic change of the user's use scene is also an important factor, for example, the simultaneous use of multiple networked devices by family members, or the frequent switching of video resources in a short time, etc., which can cause network congestion and lead to unreasonable allocation of network resources, thereby aggravating the problem of unstable connection.

[0003] The existing network optimization methods of the set-top box are mainly based on static configuration or single-point optimization, and the common means include automatic channel scanning and switching, fixed QoS priority rule setting, simple channel quality judgment based on received signal strength and signal-to-noise ratio, and the introduction of a timing temperature control mechanism on some high-end devices. Although these methods can improve the local performance in a specific scene, they generally lack the comprehensive perception ability of the dynamic environment and the running state of the device. SUMMARY

[0004] Therefore, the present application provides a network optimization method and device of a set-top box, an electronic device and a storage medium, to solve the problem of lack of multi-dimensional collaborative perception and joint decision of the network environment and the device state.

[0005] The first aspect of the present application provides a network optimization method of a set-top box, the method comprising: perception fusion is performed on the collected initial environment data set and the set-top box running data to generate environment and device joint data; channel frequency band optimization processing is performed according to the environment and device joint data to generate a channel resource allocation list; service optimization decision processing is performed according to the real-time traffic sequence of the set-top box and the channel resource allocation list to obtain a service optimization strategy; heat dissipation and power consumption collaborative control is performed according to a preset temperature threshold and real-time collected temperature change data to generate a heat management instruction set; network resource joint scheduling is performed according to the service optimization strategy and the heat management instruction set to generate a global optimization strategy.

[0006] In an optional implementation, the perception fusion of the collected initial environment data set and the set-top box running data to generate environment and device joint data includes: filtering and spatio-temporal alignment processing of the real-time collected initial environment data set to obtain physical environment data; smoothing processing of the real-time collected set-top box running data to obtain target device state data; data fusion processing of the physical environment data and the target device state data to generate environment and device joint data.

[0007] In an optional implementation, the channel frequency band optimization processing according to the environment and device joint data to generate a channel resource allocation list includes: Z-score standardization processing of the environment and device joint data to obtain standardized joint data; principal component analysis and dimension reduction processing of the standardized joint data to obtain denoising optimization data; decision analysis of the denoising optimization data by a preset interference decision model to obtain a channel occupation probability set of each channel in the current environment; channel quality assessment and optimal channel screening of each channel according to the signal strength in the standardized joint data and the channel occupation probability set to obtain an optimal channel index; band switching and channel allocation processing according to the optimal channel index and the distribution data of each channel in the target frequency band interval in the standardized joint data to generate a channel resource allocation list.

[0008] In an optional implementation, the service optimization decision processing according to the real-time traffic sequence of the set-top box and the channel resource allocation list to obtain a service optimization strategy includes: real-time collection of the real-time traffic sequence of the set-top box according to a preset time window, and aggregation and formatting processing of the real-time traffic sequence to obtain a bandwidth sequence; splicing processing of the bandwidth sequence and statistical data of historical bandwidth to obtain traffic characteristics, and bandwidth demand prediction of the traffic characteristics by a preset LSTM network to obtain bandwidth demand data; linear constraint programming according to the channel resource allocation list and the bandwidth demand data to obtain an optimal bandwidth allocation set; different service level differentiation and marking processing of the optimal bandwidth allocation set to obtain a service optimization strategy.

[0009] In an optional implementation, the heat dissipation and power consumption collaborative control according to the preset temperature threshold and the temperature change data collected in real time is configured to generate a heat management instruction set, which includes: collecting temperature change data of the set-top box in real time according to a preset time period to obtain a temperature change sequence; calculating a current power dissipation and heat dissipation power according to a power consumption index in the target device state data, the temperature change sequence, and the current environment temperature collected; performing linear mapping processing on the instantaneous temperature difference data collected according to the current power dissipation and heat dissipation power to obtain a fan PWM control instruction; performing dynamic adjustment on a processor frequency according to the preset temperature threshold and the instantaneous temperature difference data to obtain a CPU target main frequency; packaging the fan PWM control instruction and the CPU target main frequency according to a preset instruction format to generate the heat management instruction set.

[0010] In an optional implementation, the global optimization strategy includes network switching instructions and resource reallocation data, and the network resource joint scheduling according to the service optimization strategy and the heat management instruction set is configured to generate a global optimization strategy, which includes: obtaining online instantaneous state data of each standby network interface, and constructing a reinforcement learning state set according to the service optimization strategy, the heat management instruction set, and the online instantaneous state data; performing optimal scheduling action screening processing on the reinforcement learning state set according to a preset Q learning model to obtain an optimal adjustment measure; performing network switching and resource allocation analysis and instruction conversion on the optimal adjustment measure to generate the network switching instructions and the resource reallocation data.

[0011] In an optional implementation, the method further includes: performing bandwidth deviation calculation according to a current bandwidth usage rate monitored in real time and a service allocation bandwidth value in the resource reallocation data to obtain bandwidth allocation feedback data; performing smoothing processing on set-top box running index data collected in real time according to a preset update algorithm and a smoothing coefficient set to update a CPU occupancy rate, a memory fragmentation rate, and a packet loss rate; updating the target device state data according to the bandwidth allocation feedback data, the CPU occupancy rate, the memory fragmentation rate, and the packet loss rate.

[0012] The second aspect of the present application provides a network optimization device of a set-top box, which includes: A multi-dimensional perception module is configured to perform perception fusion on the collected initial environment data set and the set-top box operation data to generate environment and device joint data. A signal optimization module is configured to perform channel frequency band optimization processing according to the environment and device joint data to generate a channel resource allocation list A service optimization module is configured to perform service optimization decision processing according to the real-time traffic sequence of the set-top box and the channel resource allocation list to obtain a service optimization strategy. A heat dissipation planning module is configured to perform heat dissipation and power consumption collaborative control according to a preset temperature threshold and real-time collected temperature change data to generate a heat management instruction set. A global optimization module is configured to perform network resource joint scheduling according to the service optimization strategy and the heat management instruction set to generate a global optimization strategy.

[0013] The third aspect of the present application provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the network optimization method of the set-top box when executing the computer program.

[0014] The fourth aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the steps of the network optimization method of the set-top box.

[0015] In summary, the present application has at least the following beneficial technical effects: 1. The joint and collaborative perception of environment and device state improves the accuracy of optimization decision by comprehensively considering temperature and humidity, surrounding signal interference, device operation state and other information.

[0016] 2. The interference decision model performs synchronous optimization decision on the channel and frequency band, effectively avoiding channel conflict, signal blocking and other problems caused by fixed configuration, and is particularly suitable for network dynamic optimization in high-density wireless environment.

[0017] 3. The LSTM network predicts peak traffic and burst demand to realize intelligent scheduling and priority allocation of video, voice, data and other multi-service traffic, and guarantees the smoothness and response speed of users in different use scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0019] Figure 1 is a flow chart of a network optimization method of a set top box provided by an embodiment of the present application; Figure 2 is a functional module diagram of a network optimization device of a set top box provided by an embodiment of the present application; Figure 3 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0021] As shown in Figure 1 , a flow chart of a network optimization method of a set top box provided by an embodiment of the present application. The network optimization method of a set top box provided by the embodiment of the present application includes the following steps.

[0022] Step S1, perceptual fusion is performed on the collected initial environment data set and set top box running data to generate environment and device joint data.

[0023] It should be understood that the embodiment of the present application adopts the process of optimizing the network of a set top box in a network environment, and the network optimization method of a set top box of the present application is explained and described. There is also a connected distributed sensor network in the network environment mentioned in the embodiment of the present application, which includes but is not limited to: an environmental temperature perception array, a wireless spectrum probe, a smart power monitoring device, and a device state acquisition module integrated in a set top box (for example, a CPU occupation monitor, a memory state tracking module, a network packet statistics module). Each module is periodically polled by an electronic device executing the network optimization method of a set top box described in the embodiment of the present application through an acquisition program to generate an original multi-source data stream. The original multi-source data stream includes but is not limited to an initial environment data set and set top box running data.

[0024] Specifically, the initial environment dataset includes temperature, wireless signal strength and voltage fluctuation data. For the collected initial environment dataset, the noise characteristics of each sensor data in time and space are first considered, and therefore when the original temperature and frequency spectrum readings are obtained, the embodiments of the present application apply a multi-scale filter based on adaptive noise estimation to smooth the irregular mutations of each sensing point into a usable environment state sequence. The filter automatically adjusts the filter coefficients in different time windows to balance the fast response to sudden interference and the smooth tracking of long-term trends, so as to provide a set of high-confidence temperature, signal power and voltage variance data. For example, when the router of a neighbor house in a residential building suddenly changes the channel, causing signal fluctuation, the filter can identify and eliminate abnormal peaks within a few hundred milliseconds, thereby avoiding subsequent strategy misjudgment. Subsequently, the readings of all sensing points are unified to the same reference time and spatial grid through a time-space alignment algorithm, which maps the measurement sequences scattered in different coordinates and millisecond-level delays to the same two-dimensional plane and synchronous time through interpolation and time offset compensation, so as to ensure that the subsequent fusion engine can align and compare the environmental intensity differences of each region and each time.

[0025] At the same time, the set-top box running data includes CPU occupancy rate, memory fragmentation rate and network packet loss rate. After obtaining the set-top box running data, it is necessary to smooth the set-top box running data by a dynamic weighted moving average method, which automatically adjusts the weight distribution of new and old data according to the current load fluctuation amplitude, so as to ensure that the slowly changing part is captured through a long memory window, and the sudden fluctuation part is quickly responded through a short memory window. For example, when the CPU occupancy suddenly rises due to the start of high-definition video decoding, this weighted moving average can quickly reflect the change within a limited period, and will not cause system misjudgment due to accidental jitter, thereby outputting a set of state vectors reflecting the real set-top box running trend.

[0026] Finally, in order to unify and merge the physical environment data and target device state data obtained after the above processing, the embodiment of the application introduces a fusion engine with cross-domain correlation and nonlinear mapping capabilities. First, the obtained physical environment data and the smoothed target device state data are input into the fusion engine in a weighted combination manner. The fusion engine maps the combination result through a nonlinear activation function, thereby generating a high-dimensional joint fingerprint (i.e., environment and device joint data) that can depict both the environment interference intensity and the device running state. The weighted combination process weights and sums the environment and device information according to a pre-set weight coefficient, and suppresses negative effects through a linear rectifier unit, ensuring that the final data has both stability and acuteness when reflecting complex interactions. For example, in a scenario where the environment temperature rises and the packet loss rate also rises synchronously, the fusion engine can strengthen the superimposed effect of the two, so that the subsequent channel selection and bandwidth scheduling strategy preferentially avoids the disturbed area. This joint data not only provides a unified and reliable input benchmark for subsequent modules, but also ensures that the system can make fine resource optimization decisions in a timely manner when the network environment changes dynamically.

[0027] The generated environment and device joint data also needs to be further subjected to small batch normalization to make it uniformly distributed in each dimension, facilitating subsequent decision model processing. For example, when the environment temperature rises while the packet loss rate also increases, the bilinear interaction amplifies the response of this combined dimension, so that the fusion result can directly drive the subsequent channel selection module to preferentially avoid the most severely disturbed frequency band. Filtering alignment ensures data reliability, and weighted smoothing ensures accurate running state.

[0028] Step S2, performing channel frequency optimization processing according to the environment and device joint data to generate a channel resource allocation list.

[0029] First, the environment and device joint data output from the fusion engine is input into the standardization module in the form of a vector. The Z-score standardization algorithm is used to subtract the historical average of each dimension from the numerical value and divide it by the historical standard deviation, obtaining standardized joint data with zero mean and unit variance, thereby eliminating the influence of different dimensions and orders of magnitude on subsequent analysis. This processing can ensure stable feature distribution when network signals and device states fluctuate extremely, for example, when CPU occupancy drops sharply and signal strength increases sharply in energy saving mode, standardization can prevent any single indicator from dominating the entire decision-making process.

[0030] Then, principal component analysis technology is used to reduce the dimension of the standardized joint data, and the latitude and longitude features are mapped into a principal component space with the highest degree of interpretation to retain more than 90% of the data information and significantly reduce the model calculation amount. By calculating the eigenvectors of the covariance matrix and retaining a number of principal components with a cumulative variance of 90%, the high-dimensional data is mapped into a low-dimensional space, which not only retains the core information of the interaction between the environment and the device, but also greatly compresses the model input dimension. This process can avoid performance bottlenecks when the input dimension of the channel evaluation model is too high, while maintaining the sensitive capture of the interaction features of the environment and the device state. Then, the de-noising and optimized data after dimension reduction are input into the pre-trained deep interference decision model for decision analysis. The model based on the deep residual network structure can automatically extract the nonlinear cross features in the data and output the occupation probability vector of each channel to measure the utilization priority of each channel in the current environment.

[0031] After obtaining the channel occupation probability, it is necessary to quantitatively evaluate each channel in combination with the signal and interference strength. The embodiment of the present application quantitatively evaluates the channel quality, so that the channel selection matches the prediction model preference and also takes into account the actual transmission capacity, and needs to fuse the channel occupation probability and the channel capacity characteristics to construct the evaluation logic. First, according to the model output of the channel occupation probability, in combination with the signal strength, noise and interference data (for example, noise power spectral density, intensity of each interference source) reflecting the actual transmission potential of the channel, the logarithmic term related to the channel capacity (i.e., the signal, noise and interference relationship is used to measure the basic transmission capacity of the channel) is calculated, and then the capacity term is related to the power of the channel occupation probability, and the probability preference and capacity potential are integrated through nonlinear transformation, and finally the channel quality score Q i is output. The channel quality score Q i takes into account the nonlinear influence of probability and capacity, so that the channel selection considers the prediction model preference and also takes into account the actual transmission capacity.

[0032] After calculating the Q i value of all channels, the channel index with the maximum Q i value is selected, and then a specific frequency band switching instruction is generated. The instruction is issued to the home gateway in the form of a control message, instructing it to switch the main communication channel to the selected frequency band. At the same time, in combination with the distribution information of each channel in the standardized joint data among different frequency bands, a channel resource allocation list is constructed, which lists the bandwidth range and priority order of each channel available in the future, so that in the next round of network optimization cycle, the adjacent frequency interference channels are enabled or disabled in the mapping order.

[0033] Step S3, performing service optimization decision processing according to the real-time traffic sequence of the set-top box and the channel resource allocation list to obtain a service optimization strategy.

[0034] After obtaining the channel resource allocation list, the network packet header information of the set-top box is captured in real time in each half-second time window, and the bandwidth usage sequence of each service is aggregated according to the time window, so that the original packet length and arrival timestamp are converted into a directly quantifiable bandwidth value sequence. In this process, all packets are divided into continuous time periods, and the total byte number of each period is calculated and then divided by the time window length to obtain the bandwidth throughput index in unit time, so that the bandwidth demand fluctuation can be immediately reflected when the video playback bursts or the file download rate suddenly rises. Then, the generated bandwidth sequence is connected with the historically stored bandwidth statistical data in the time dimension to construct a time sequence feature vector (i.e., traffic feature) with context memory, which is input into the LSTM network that has been trained in advance. The LSTM network is a recurrent neural network that can capture long sequence dependencies, and through its internal gating mechanism, information is screened and retained, and finally the bandwidth demand data of each service in the next ten seconds is output. This prediction can help to distinguish between occasional peaks and sustained high demand, such as when the user switches to a super-clear live channel, the prediction model can adapt to higher bandwidth requirements within a few seconds.

[0035] Subsequently, the bandwidth demand data is combined with the corresponding total channel capacity in the channel resource allocation list. In order to allocate limited resources fairly while meeting the predicted demand of all services, the present embodiment adopts the following allocation formula based on power weight: where B pred (i) is the bandwidth prediction value of the i-th service. λ is the power coefficient for controlling the balance between fairness and efficiency, which is used to suppress the extreme influence of excessively high or low prediction values on the allocation result. C total is the current total bandwidth available to the mapping list. Through this formula, the subjective demand can be met while avoiding the exclusive occupation of resources by a single service. For example, when video and download services simultaneously demand a surge, the power coefficient can effectively balance the allocation ratio of the two services and prevent the download service from completely squeezing the video streaming bandwidth.

[0036] After calculating the allocation bandwidth of each service, the optimal bandwidth allocation set is mapped to the corresponding differentiated services code point (DSCP) label, and each label corresponds to a different priority level to control the packet queue scheduling in the network. The service level determination rule is pre-configured in the controller, such as higher allocation bandwidth services are assigned higher priority, so that they can still enjoy sufficient bandwidth guarantee when the network is congested. The final output of the service optimization strategy includes the DSCP label and bandwidth quota information of each service, which can be directly issued by the SDN controller to the home gateway, providing an access threshold and execution benchmark for the next round of network optimization.

[0037] Step S4, according to the preset temperature threshold and the real-time collected temperature change data, heat dissipation and power consumption collaborative control is performed to generate a set of thermal management instructions.

[0038] For the first time in a predetermined sampling period, the system built-in temperature sensor will continuously record the temperature change curve of the key components of the set-top box, realize the real-time collection of the average temperature, peak temperature difference and temperature rise rate, and this process helps to capture the thermal response of the processor under high load scenarios (such as long time high-definition video decoding). After the collection is completed, the power consumption indicators in the target device state data, that is, the chip capacitance switching power consumption and static leakage power consumption, are combined to construct a nonlinear thermal balance model to accurately predict the heat dissipation rate of the current system. The model uses an innovative power relationship to describe the nonlinear coupling between temperature difference and heat dissipation efficiency, and its key formula is: 散 P η = H (ΔT) 2 + Γ (CV σ f) 散 . Where P 散 represents the total power dissipated by the system. H is the environmental convective heat transfer coefficient. ΔT represents the difference between the internal chip temperature of the set-top box and the ambient temperature. η is used to amplify the improvement effect of heat dissipation efficiency under high temperature difference conditions. Γ is the power consumption weighting coefficient. C is the chip capacitance. V is the current working voltage. f is the current CPU main frequency. σ is used to balance the nonlinear influence of dynamic power consumption on heat contribution. By inputting the temperature difference and real-time power consumption data, the total heat dissipation capacity is calculated, thereby guiding the subsequent fan speed and frequency reduction decisions. In the model solving stage, the current environmental temperature and the collected temperature change sequence are brought into the formula, and the current power dissipation P 散 is obtained by numerical iteration method. In the iteration process, the time-varying characteristics of internal and external temperature difference and power consumption fluctuation are considered in each calculation to ensure that the new balance value can be quickly converged when a sudden high load starts. After completing the calculation of heat dissipation power, the linear mapping technology is applied to the instantaneous temperature difference data to map the temperature difference to the fan duty cycle level. The mapping relationship is defined as the ratio of duty cycle to heat dissipation power (i.e., fan PWM control instruction), for example, when the temperature difference rises to 15℃ and P 目标 reaches 8W, the duty cycle will automatically adjust to the ratio of 200 / 255 to ensure that the fan speed is sufficient to remove excess heat from the case.

[0039] At the same time, combined with the preset temperature threshold (for example, the upper limit of safe operation is 70℃), the system will dynamically adjust the current main frequency of the processor. If the measured instantaneous temperature difference exceeds the threshold for more than a few seconds, the system will perform an exponential decreasing algorithm to gradually reduce the main frequency according to the temperature difference and power consumption ratio to avoid overheating leading to hardware degradation or damage. The formula can be expressed as f max×exp(-αΔT), where α is the temperature difference inhibition coefficient (about 0.02), which can quickly reduce the frequency to control power consumption and temperature rise when the temperature difference increases significantly.

[0040] Finally, the fan PWM control instruction and the CPU target main frequency are encapsulated into a standardized control message according to a preset communication protocol, and are sent to the fan driver and the power management unit through the communication bus. This instruction set not only can timely improve the heat dissipation capacity in high temperature working conditions, but also can smoothly adjust the main frequency and the fan speed under ordinary load, so as to balance the heat dissipation effect and the running noise. Through deep coupling modeling and real-time solving of the temperature change sequence, the equipment power consumption state and the environment temperature, the thermal management instruction set realizes fine collaborative control of the set-top box heat dissipation and power consumption, and guarantees stable and reliable operation of the system in various use scenarios.

[0041] Step S5, performing network resource joint scheduling according to the service optimization strategy and the thermal management instruction set to generate a global optimization strategy.

[0042] The primary work of fusing the quality of service rules and the thermal management instructions in the multi-network environment is to perform online state detection on each standby network interface. The network monitoring module collects the connectivity, round-trip delay and packet loss rate of each interface in real time, and integrates this information with the previously generated service optimization strategy and the thermal management instruction set into a high-dimensional state vector (i.e., a reinforcement learning state set). The reinforcement learning state set not only contains the priority marks and bandwidth quotas of each service, but also includes the fan speed instructions and the processor main frequency thresholds, and further injects the availability index of the network interface, so as to condense the three dimensions of traffic requirements, thermal safety demands and network reachability in a unified representation space, and provide a global view for subsequent intelligent scheduling.

[0043] After the state vector is constructed, the reinforcement learning module continuously updates the decision strategy by using the built-in Q learning model. The Q learning model used in the embodiment of the application can be expressed by the following formula: wherein, represents the total bandwidth actually obtained and effectively used by each service under the current action. max is the maximum theoretical bandwidth of the channel. α is used to balance the weight of throughput efficiency and the last two terms. inst and T env are the current temperature of the set-top box chip and the environment temperature, respectively. thresh is the safety temperature threshold. β determines the sensitivity of the heat preservation margin to the decision. avail and L totalrespectively. γ is used to strengthen the utilization of multi-network redundancy. The core reward function of the Q-learning model is not simply the cumulative business throughput or temperature penalty, but a kind of multiplicative coupling method. It can amplify the reward in the ideal state of high bandwidth, high residual temperature difference and multi-link availability, and quickly decay in the case of any deficiency, in order to drive the decision while taking into account throughput, heat dissipation and network reliability.

[0044] The model is updated for the current state s and action a using the reward function, and the Q table is corrected online by maximizing the Bellman optimality principle of future cumulative reward. The model will adaptively select the optimal action (i.e., the optimal adjustment measure) in each control cycle. The optimal action set selected from the Q table often contains network switching operations (such as enabling or disabling a specific interface) and resource reallocation schemes (such as increasing the bandwidth of a certain service, degrading non-critical traffic, etc.).

[0045] Subsequently, the system parses the selected optimal adjustment measure into two parts: network switching instructions and resource reallocation data. The network switching instructions are in the form of SDN control messages and are sent to the home gateway, instructing it to turn on or off a routing link or perform switching between primary and backup channels. The resource reallocation data lists the new bandwidth allocation for each service and the corresponding service priority, which is delivered to the flow control module in a tabular data structure for immediate effect. For example, when online video and cloud backup are simultaneously high load, the model may choose to switch to a low-latency Wi-Fi6 link to ensure smooth video streaming, and further increase the bandwidth allocation mapping matrix for high-definition video, while reducing the priority of backup tasks.

[0046] Through the above high-dimensional state integration, innovative multiplicative reward function and Q-learning online update, combined with the conversion mechanism of the parsed action into network and resource atomic instructions, the multi-network fusion scheduling of anti-heat and performance preservation, bandwidth preservation and reliability enhancement is realized, forming an end-to-end closed-loop optimization, which significantly improves the network performance and operational stability of the home set-top box.

[0047] In an optional implementation, the execution effect of the previous round of resource scheduling is fed back to the system to realize a closed-loop adaptive optimization of the whole process. After the resource reallocation data is issued and starts to execute, the actual bandwidth usage of each service is monitored, and the deviation between the actual bandwidth usage and the bandwidth value allocated to each service in the mapping matrix is calculated to quantify the allocation accuracy. For this purpose, the real-time bandwidth utilization ratio and the allocation ratio of each service are regarded as two probability distributions, and the difference between the two is measured by relative entropy as shown in the following formula: where B used (i) and B alloc(i) respectively represent the real-time usage bandwidth and the allocated bandwidth of the i-th service. The ratio of the sum of the numerator and denominator is used to normalize the probability distribution, and the greater the divergence indicates the more significant the difference between allocation and usage. For example, if the online video service is allocated 50 Mbps but actually only uses 10 Mbps, the relative entropy of this contribution will be higher, prompting the subsequent scheduling to tighten the bandwidth share or adjust the priority of this service to avoid resource waste.

[0048] After obtaining the bandwidth deviation feedback data, the influence of the bandwidth difference on the system operation index needs to be smoothed to avoid unnecessary drastic fluctuations in configuration caused by short-term jitter. The exponential smoothing algorithm based on deviation driving shown below is used in the embodiments of the present application to combine the currently collected CPU occupancy rate, memory fragmentation rate and packet loss rate with the last smoothing value: wherein S can replace any of the performance indicators of CPU occupancy rate, memory fragmentation rate and packet loss rate. S int represents the real-time acquisition value. S old is the last smoothing result. θ is a positive coefficient for controlling the smoothing strength. The greater the relative entropy, the closer the weight to 1, which means that the system dynamics are more sensitive to the deviation change; when the deviation is small, more historical data is retained to suppress short-term jitter. For example, when the network packet loss rate temporarily increases due to sudden route switching, if the duration of this condition is short, the smoothing factor can effectively prevent it from being equated to a system anomaly, thereby avoiding frequent adjustment of channel or power consumption strategies.

[0049] Finally, the above bandwidth deviation feedback data and the smoothed CPU occupancy rate, memory fragmentation rate and packet loss rate are written into the target device state data to replace the original corresponding fields, thereby updating the running side of the set-top box in the current environment. The new matrix not only reflects the actual bandwidth utilization effect after resource scheduling is executed, but also integrates the smoothing result of the device running trend, providing a more accurate basis for subsequent regeneration of environment and device joint data. For example, when multiple users are online gaming and streaming media at the same time, if the bandwidth allocation deviation is large and the CPU occupancy rate continues to rise, the system will increase the sensitivity to the computational load fluctuation in the next round of fusion, further driving the optimization of channel selection and thermal management decisions, achieving true closed-loop adaptive control.

[0050] The application belongs to the technical field of Internet, and generates environment and device joint data by sensing fusion of initial environment data set and set top box running data, generates channel resource allocation list by channel frequency optimization processing according to the environment and device joint data, obtains service optimization strategy by service optimization decision processing according to real-time traffic sequence of the set top box and the channel resource allocation list, generates heat management instruction set by heat dissipation and power consumption collaborative control according to the temperature threshold and the real-time collected temperature change data, and generates global optimization strategy by network resource joint scheduling according to the service optimization strategy and the heat management instruction set. The application improves the network performance and response speed of the set top box by multi-source information fusion, AI decision modeling, heat management control and cross-domain scheduling mechanism, and strengthens the energy consumption control and stability guarantee of the device level, which has important practical value for constructing intelligent home terminal network.

[0051] As shown in Figure 2 FIG. 1 is a functional module diagram of a network optimization device of a set top box provided by an embodiment of the application.

[0052] In some embodiments, the network optimization device 2 of the set top box can include a plurality of functional modules composed of computer program segments. The computer programs of each program segment in the network optimization device 2 of the set top box can be stored in the memory of the server and executed by at least one processor to perform the functions of the network optimization method of the set top box (see Figure 1 for detailed description).

[0053] In this embodiment, the network optimization device 2 of the set top box can be divided into a plurality of functional modules according to its functions. The functional modules can include a multi-dimensional sensing module 21, a signal optimization module 22, a service optimization module 23, a heat dissipation planning module 24, a global optimization module 25 and an execution feedback module 26. The module referred to by the present application refers to a series of computer program segments that can be executed by at least one processor and can complete a fixed function, which are stored in the memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0054] The multi-dimensional sensing module 21 is configured to perform sensing fusion on the collected initial environment data set and set top box running data to generate environment and device joint data.

[0055] In an optional embodiment, the multi-dimensional sensing module 21 is specifically configured to: perform filtering and space-time alignment processing on the real-time collected initial environment data set to obtain physical environment data; perform smoothing processing on the real-time collected set top box running data to obtain target device state data; perform data fusion processing on the physical environment data and the target device state data to generate environment and device joint data.

[0056] a signal optimization module 22 configured to perform channel band optimization processing on the environment and device joint data to generate a channel resource allocation list.

[0057] In an optional embodiment, the signal optimization module 22 is specifically configured to: perform Z-score standardization processing on the environment and device joint data to obtain standardized joint data; perform principal component analysis and dimension reduction processing on the standardized joint data to obtain denoised optimization data; perform decision analysis on the denoised optimization data through a preset interference decision model to obtain a channel occupation probability set of each channel in the current environment; perform channel quality assessment and optimal channel screening on each channel according to a signal strength in the standardized joint data and the channel occupation probability set to obtain an optimal channel index; perform band switching and channel allocation processing according to the optimal channel index and distribution data of each channel in a target band interval in the standardized joint data to generate a channel resource allocation list.

[0058] a service optimization module 23 configured to perform service optimization decision processing according to a real-time traffic sequence of a set-top box and the channel resource allocation list to obtain a service optimization strategy.

[0059] In an optional embodiment, the service optimization module 23 is specifically configured to: collect a real-time traffic sequence of a set-top box in real time according to a preset time window, and perform aggregation and formatting processing on the real-time traffic sequence to obtain a bandwidth sequence; perform splicing processing on the bandwidth sequence and statistical data of historical bandwidth to obtain traffic features, and perform bandwidth demand prediction on the traffic features through a preset LSTM network to obtain bandwidth demand data; perform linear constraint programming according to the channel resource allocation list and the bandwidth demand data to obtain an optimal bandwidth allocation set; perform different service level differentiation and marking processing on the optimal bandwidth allocation set to obtain a service optimization strategy.

[0060] a heat dissipation planning module 24 configured to perform heat dissipation and power consumption collaborative control according to a preset temperature threshold and temperature change data collected in real time to generate a heat management instruction set.

[0061] In an optional embodiment, the heat dissipation planning module 24 is specifically configured to: acquire temperature change data of the set-top box in real time according to a preset time period to obtain a temperature change sequence; calculate a current power dissipation cooling power according to a power consumption index in the target device state data, the temperature change sequence, and the acquired current ambient temperature; perform linear mapping processing on the acquired instantaneous temperature difference data according to the current power dissipation cooling power to obtain a fan PWM control instruction; perform processor frequency dynamic adjustment according to a preset temperature threshold and the instantaneous temperature difference data to obtain a CPU target main frequency; package the fan PWM control instruction and the CPU target main frequency according to a preset instruction format to generate a thermal management instruction set.

[0062] The global optimization module 25 is configured to perform network resource joint scheduling according to the service optimization strategy and the thermal management instruction set to generate a global optimization strategy.

[0063] In an optional implementation, the global optimization module 25 is specifically configured to: acquire online instantaneous state data of each standby network interface, and construct a reinforcement learning state set according to the service optimization strategy, the thermal management instruction set, and the online instantaneous state data; perform optimal scheduling action screening processing on the reinforcement learning state set according to a preset Q learning model to obtain an optimal adjustment measure; perform network switching and resource allocation analysis and instruction conversion on the optimal adjustment measure to generate the network switching instruction and the resource reallocation data.

[0064] In an optional implementation, the network optimization apparatus 2 of the set-top box further includes an execution feedback module 26, which is specifically configured to: perform bandwidth deviation calculation according to a current bandwidth usage rate monitored in real time and a service allocation bandwidth value in the resource reallocation data to obtain bandwidth allocation feedback data; perform smoothing processing on set-top box running index data acquired in real time according to a preset update algorithm and a smoothing coefficient set to update a CPU occupancy rate, a memory fragmentation rate, and a packet loss rate; update the target device state data according to the bandwidth allocation feedback data, the CPU occupancy rate, the memory fragmentation rate, and the packet loss rate.

[0065] It should be understood that the various changes and specific embodiments of the method provided by the above embodiments are also applicable to the network optimization device of the set top box of the present embodiment. Through the foregoing detailed description of the network optimization method of the set top box, those skilled in the art can clearly understand the implementation method of the network optimization device of the set top box in the present embodiment. For the sake of brevity of the description, it will not be described in detail here.

[0066] As shown in Figure 3 , it is a structural schematic diagram of an electronic device provided by the embodiment of the present application.

[0067] In the preferred embodiment of the present application, the electronic device 3 can include, but is not limited to, a memory 31, at least one processor 32, and at least one communication bus 33.

[0068] Those skilled in the art should understand that Figure 3 The structure of the electronic device 3 shown is not a limitation of the embodiment of the present application. The electronic device 3 can also include more or less other hardware or software, or different component arrangements.

[0069] In some embodiments, the electronic device 3 is a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored instructions. The hardware thereof includes, but is not limited to, a microprocessor, an application specific integrated circuit, a programmable gate array, a digital processor, and an embedded device, etc.

[0070] It should be noted that the electronic device 3 is only an example. Other existing or future electronic products, such as those that can be adapted to the present application, should also be included within the protection scope of the present application and are hereby incorporated by reference.

[0071] In some embodiments, the memory 31 stores a computer program which, when executed by the at least one processor 32, implements all or part of the steps of the network optimization method of the set-top box as described. The memory 31 includes a Read-Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), a One-time Programmable Read-Only Memory (OTPROM), an Electrically-Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM), or other optical disk storage, magnetic disk storage, magnetic tape storage, or any other medium of storage of computer-readable data now known or later developed. Further, the computer-readable storage medium can include a program storage area and a data storage area, wherein the program storage area can store an operating system, application programs required by at least one function, etc.

[0072] In some embodiments, the at least one processor 32 is a control unit of the electronic device 3, which connects various components of the entire electronic device 3 through various interfaces and lines, and performs various functions of the electronic device 3 and processes data by running or executing programs or modules stored in the memory 31 and calling data stored in the memory 31. For example, the at least one processor 32 implements all or part of the steps of the network optimization method of the set-top box as described in the embodiments of the present application when executing the computer program stored in the memory 31, or implements all or part of the functions of the network optimization apparatus of the set-top box. The at least one processor 32 can be composed of integrated circuits, for example, can be composed of a single packaged integrated circuit, or can be composed of multiple packaged integrated circuits with the same function or different functions, including one or more combinations of a Central Processing Unit (CPU), a microprocessor, a digital processing chip, a graphics processor, and various control chips, etc.

[0073] In some embodiments, the at least one communication bus 33 is configured to enable connection communication between the memory 31 and the at least one processor 32. Although not shown, the electronic device 3 can further include a power supply (such as a battery) for powering the various components of the electronic device 3. Preferably, the power supply is logically connected to the at least one processor 32 via a power management device, thereby enabling management of charging, discharging, and power consumption management, etc. by the power management device. The power supply can also include one or more direct current or alternating current power sources, recharging circuits, power failure detection circuitry, power converters or inverters, power status indicators, and the like. The electronic device 3 can further include a variety of sensors, a Bluetooth module, a Wi-Fi module, and the like, which are not described herein.

[0074] The integrated units in the form of software function modules described above can be stored in a computer readable storage medium. The software function modules described above are stored in a storage medium, and include a plurality of instructions for causing an electronic device (which can be a personal computer, an electronic device, or a network device, etc.) or a processor to execute part of the method described in various embodiments of the present application.

[0075] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the described apparatus embodiments are merely illustrative. For example, the division of the modules is merely a logical function division. There can be another division manner in actual implementation.

[0076] The modules illustrated as separate components can or can not be physically separate, and the components illustrated as modules can or can not be physical units. They can be located in one place, or distributed on a plurality of network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiments.

[0077] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Therefore, any equivalent changes made on the structure, shape, and principle of the present application should be covered within the protection scope of the present application.

Claims

1. A network optimization method of a set-top box, characterized by, The method comprises: performing perception fusion on the collected initial environment data set and set-top box operation data to generate environment and device joint data; performing channel frequency band optimization processing according to the environment and device joint data to generate a channel resource allocation list; performing service optimization decision processing according to the real-time traffic sequence of the set-top box and the channel resource allocation list to obtain a service optimization strategy; performing heat dissipation and power consumption collaborative control according to a preset temperature threshold and real-time collected temperature change data to generate a heat management instruction set; performing network resource joint scheduling according to the service optimization strategy and the heat management instruction set to generate a global optimization strategy.

2. The network optimization method of the set-top box according to claim 1, wherein, The perception fusion on the collected initial environment data set and set-top box operation data to generate environment and device joint data comprises: performing filtering and space-time alignment processing on the real-time collected initial environment data set to obtain physical environment data; performing smoothing processing on the real-time collected set-top box operation data to obtain target device state data; performing data fusion processing on the physical environment data and the target device state data to generate environment and device joint data.

3. The network optimization method of set-top box according to claim 1, wherein, The channel frequency band optimization processing according to the environment and device joint data to generate a channel resource allocation list comprises: performing Z-score standardization processing on the environment and device joint data to obtain standardized joint data; performing principal component analysis and dimension reduction processing on the standardized joint data to obtain denoising optimization data; performing decision analysis on the denoising optimization data through a preset interference decision model to obtain a channel occupation probability set of each channel in the current environment; performing channel quality assessment and optimal channel screening on each channel according to the signal strength in the standardized joint data and the channel occupation probability set to obtain an optimal channel index; performing frequency band switching and channel allocation processing according to the optimal channel index and the distribution data of each channel in the target frequency band interval in the standardized joint data to generate a channel resource allocation list.

4. The network optimization method of set-top box according to claim 1, wherein, The service optimization decision processing according to the real-time traffic sequence of the set-top box and the channel resource allocation list to obtain a service optimization strategy comprises: collecting the real-time traffic sequence of the set-top box according to a preset time window in real time, and performing aggregation and formatting processing on the real-time traffic sequence to obtain a bandwidth sequence; performing splicing processing on the bandwidth sequence and historical bandwidth statistical data to obtain traffic characteristics, and performing bandwidth demand prediction on the traffic characteristics through a preset LSTM network to obtain bandwidth demand data; performing linear constraint programming according to the channel resource allocation list and the bandwidth demand data to obtain an optimal bandwidth allocation set; performing different service level differentiation and marking processing on the optimal bandwidth allocation set to obtain a service optimization strategy.

5. The network optimization method of set-top box according to claim 2, wherein, The heat dissipation and power consumption collaborative control according to a preset temperature threshold and real-time collected temperature change data to generate a heat management instruction set comprises: collecting the temperature change data of the set-top box according to a preset time period in real time to obtain a temperature change sequence; According to the power consumption index in the target device state data, the temperature change sequence, and the current environment temperature collected, a current power consumption heat dissipation power is calculated; According to the current power consumption heat dissipation power, the real-time temperature difference data collected is linearly mapped to obtain a fan PWM control instruction; According to the preset temperature threshold and the real-time temperature difference data, a processor frequency dynamic adjustment is performed to obtain a CPU target main frequency; According to a preset instruction format, the fan PWM control instruction and the CPU target main frequency are packaged to generate a thermal management instruction set.

6. The network optimization method of set-top box according to claim 5, wherein, The global optimization strategy includes network switching instructions and resource reallocation data, and the network resource joint scheduling is performed according to the service optimization strategy and the thermal management instruction set to generate a global optimization strategy, which includes: Online real-time state data of each standby network interface is obtained, and a reinforcement learning state set is constructed according to the service optimization strategy, the thermal management instruction set, and the online real-time state data; Through a preset Q learning model, an optimal adjustment measure is obtained by performing optimal scheduling action filtering processing on the reinforcement learning state set; The optimal adjustment measure is analyzed and converted into network switching instructions and resource reallocation data.

7. The network optimization method of set-top box according to claim 6, wherein, The method further includes: According to the real-time monitored current bandwidth usage and the service allocation bandwidth value in the resource reallocation data, a bandwidth deviation is calculated to obtain bandwidth allocation feedback data; According to a preset update algorithm and a smoothing coefficient set, the CPU occupancy rate, the memory fragmentation rate, and the packet loss rate are updated by performing smoothing processing on the real-time collected set-top box running index data; The target device state data is updated according to the bandwidth allocation feedback data, the CPU occupancy rate, the memory fragmentation rate, and the packet loss rate.

8. A network optimization apparatus of a set-top box, characterized by comprising: The device includes: A multi-dimensional perception module is configured to perform perception fusion on the collected initial environment data set and the set-top box running data to generate environment and device joint data; A signal optimization module is configured to perform channel frequency optimization processing on the environment and device joint data to generate a channel resource allocation list A service optimization module is configured to perform service optimization decision processing on the real-time traffic sequence of the set-top box and the channel resource allocation list to obtain a service optimization strategy; A heat dissipation planning module is configured to perform heat dissipation and power consumption collaborative control according to a preset temperature threshold and real-time collected temperature change data to generate a thermal management instruction set; A global optimization module is configured to perform network resource joint scheduling according to the service optimization strategy and the thermal management instruction set to generate a global optimization strategy.

9. An electronic device, comprising: The electronic device includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement the steps of the set-top box network optimization method according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the set-top box network optimization method according to any one of claims 1 to 7.

Images