HomeCloud-based computing unloading architecture and unloading method

Abstract

The invention discloses a computing unloading architecture and unloading method based on home cloud HomeCloud, and relates to the field of edge cloud computing. The system comprises a local area network for communication, a mobile intelligent terminal for unloading, computing equipment for providing unloading service and a HomeCloud intelligent control center, the local area network adopts a local area network in a family to carry out data transmission communication; when the mobile intelligent terminal prepares to carry out a calculation task, task information is sent to the resource sharing middleware, and if a device capable of carrying out a calculation unloading task exists, the task is unloaded to the calculation device for processing under the assistance of the resource sharing middleware; the computing device comprises an existing computing device and a computing device special for HomeCloud computing unloading. According to the method, the existing computing equipment is deeply utilized, so that the equipment which is still normally used can be fully utilized, and the utilization rate of the existing computing resources can be greatly improved.

Description

Technical Field

[0001] The present invention belongs to the field of edge cloud computing, and in particular relates to a computing offloading architecture and an offloading method based on a home cloud HomeCloud. Background Art

[0002] Due to the popularity of smart terminals and the increase in network edge nodes, the number of computing tasks that need to be processed at the network edge nodes has increased. Processing data at the network edge nodes will increase the energy consumption of the device itself and improve the performance requirements of the device. Edge computing offloading technology migrates tasks that originally need to be completed on the user's personal device to the edge server for processing. The device itself does not need to perform a lot of calculations, but only needs to pay the energy cost required for network transmission. It is particularly suitable for computing-intensive tasks. Such tasks transmit less data but require a lot of calculations to complete. The smaller transmission volume saves the communication transmission time of the task. The processing of a large number of computing tasks by the edge server not only saves the energy consumption of the user's device, but also reduces the processing time of the task and reduces the delay because the edge server has stronger computing power and faster task processing speed.

[0003] With the further development of VR head displays, handheld devices, etc., the number of smart mobile terminals used in user homes has further increased. These devices have high energy consumption. In order to extend their battery life, larger and heavier batteries must be used to increase their total power load, but this method reduces their portability and affects the user experience. Therefore, the present invention hopes to use computing offloading technology to offload computing tasks on smart mobile terminals, thereby reducing the energy consumption of smart mobile terminals, extending their battery life, and giving smart mobile terminals the ability to handle larger computing tasks. However, if the existing edge computing offloading method is used to deal with this problem, due to the large amount of computing tasks of smart mobile terminals, the workload of edge servers will increase significantly, and the requirements for the number and computing power of edge servers will also increase further, thereby increasing the cost of edge server construction. At the same time, there are a large number of powered computing devices such as computers in user homes; when users are using smart mobile terminals, these devices in their homes are likely to be idle, resulting in a large waste of computing power. The massive growth in the number of electronic products has also resulted in a large amount of electronic waste, many of which still have computing power. The generation of a large amount of electronic waste not only wastes industrial manufacturing capacity, but also significantly increases the environmental burden. How to dispose of these long-term idle electronic products is becoming an increasingly important environmental issue. Therefore, in response to the above problems, it is of great practical significance to provide a computing offload architecture and offload method based on HomeCloud. Summary of the invention

[0004] The present invention provides a computing offloading architecture and offloading method based on the home cloud HomeCloud, which utilizes existing computing devices in the home for computing offloading, including devices that are still in normal use and devices that have been idle for a long time but still retain computing capabilities. By deeply utilizing the existing computing devices, the devices that are still in normal use can be fully utilized, and the computing devices that have been idle for a long time can also perform auxiliary operations. Therefore, this method can greatly improve the utilization rate of existing computing resources, and at the same time, the reuse of long-term idle devices can also reduce the generation of electronic waste, thereby reducing the environmental burden; and this solution can also offload computing tasks on mobile smart terminals with limited power storage to the home cloud for computing, thereby saving the computing overhead of the mobile terminal and extending its battery life; computing offloading through the home cloud can reuse idle computing resources and improve the utilization rate of computing resources; in summary, the problems in the background technology are solved.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] The computing offloading architecture based on the home cloud HomeCloud of the present invention includes a local area network for communication, a mobile intelligent terminal for offloading, a computing device providing offloading services, and a HomeCloud intelligent control center;

[0007] The local area network uses a local area network within the home for data transmission and communication;

[0008] When the mobile intelligent terminal is preparing to perform a computing task, it sends the task information to the resource sharing middleware. If there is a device that can perform the computing offload task, the task is offloaded to the computing device for processing via the resource sharing middleware.

[0009] The computing devices include existing computing devices and computing devices dedicated to HomeCloud computing offloading; the existing computing devices include devices that are still in normal use by users but are idle when the smart terminal makes an offloading request, and also include devices that are idle for a long time but still have computing power; the computing power of the computing devices dedicated to HomeCloud computing offloading is stronger than that of existing computing devices, and can quickly execute large computing tasks in the home;

[0010] The HomeCloud intelligent control center is responsible for overall resource scheduling within the HomeCloud computing offloading architecture. When the mobile intelligent terminal generates or changes task requirements, it first sends task information to the HomeCloud intelligent control center, including the number of data transmission bits of the task and the number of clock cycles required for the computing task. After receiving the task requirements, the HomeCloud intelligent control center synchronizes the idle status of the computing device and the CPU computing power, and determines the offloading device based on the task requirements and computing resource utilization.

[0011] The HomeCloud intelligent control center includes a signal sending and receiving unit, a storage unit and a computing unit;

[0012] The signal sending and receiving unit is used to transmit information with the smart terminal and the HomeCloud computing offloading device; when the mobile smart terminal generates or changes the task requirement, the HomeCloud smart control center receives the information from the mobile smart terminal through the signal receiving unit; the HomeCloud smart control center sends the idle status of the synchronous computing device and the request for CPU computing power through the signal sending unit, and receives the information through the signal receiving unit; after determining the offloading plan, the HomeCloud smart control center sends the offloading plan to the corresponding computing device and the smart terminal through the signal sending unit; if the smart terminal needs to forward the task through the HomeCloud smart control center, after the HomeCloud smart control center receives the task through the signal receiving unit, it sends the task to the corresponding computing offloading device through the signal sending unit; and after receiving the task return result through the signal receiving unit, it sends it to the mobile smart terminal through the signal sending unit;

[0013] The storage unit is used to store computing offloading plans, offloading device information and intermediate data; when the HomeCloud computing offloading device determines the computing offloading plan, the plan is stored in the storage unit; the stored offloading device information includes the idle status of the computing device and the CPU computing power, as well as the corresponding latest IP address, which is updated before each computing offloading plan needs to be formulated; the storage unit stores the intermediate data for formulating the computing offloading plan, the transfer files for task transfer, and the intermediate data generated by other processes; the storage unit is a separate memory, connected to a computing offloading device for storage;

[0014] The computing unit formulates a computing offloading plan and makes a decision based on whether to perform computing offloading and the steps of determining the device and method for performing computing offloading.

[0015] Furthermore, when performing task offloading, the mobile intelligent terminal chooses to offload all or part of the task, and performs collaborative computing with the HomeCloud computing device.

[0016] Furthermore, if the task demand and the idle state of the device remain unchanged, the computing offloading task will continue to be offloaded according to the offloading plan determined above; if the idle state of the computing device changes, the computing device will send information to the HomeCloud smart control center, and the resource sharing middleware will reallocate the offloading plan.

[0017] Furthermore, the signal sending and receiving unit adopts a WIFI wireless communication module to send and receive information through WIFI signals; the signal sending and receiving unit adopts a wired WIFI interface to be connected to the WIFI AP through a cable or optical fiber.

[0018] Furthermore, the computing unit is an independent CPU computing unit or is connected to a computing offloading device, and the computing offloading device formulates a computing offloading plan.

[0019] Furthermore, the computing unit is an intelligent terminal, and the computing offloading plan is formulated directly on the intelligent terminal.

[0020] The unloading method based on the computing unloading architecture of the home cloud HomeCloud, using the computing unloading architecture based on the home cloud HomeCloud to unload, includes the following steps:

[0021] S1. The mobile smart terminal sends its uninstallation request to the HomeCloud smart control center;

[0022] S2. After receiving the task requirement, the HomeCloud intelligent control center synchronizes the CPU operation status of all computing devices, determines the computing device to perform the offloading task according to whether to perform the computing offloading and the method of determining the computing offloading device, and sends the information of the computing offloading plan and the computing offloading task to the computing device and mobile intelligent terminal corresponding to the computing offloading;

[0023] S3. The mobile smart terminal communicates directly with the computing device, or performs auxiliary communication through the HomeCloud smart control center, sends the task requirements to the computing device, and the computing device performs calculations and then returns the results to the mobile smart terminal.

[0024] Furthermore, whether to perform calculation offloading is specifically determined in the following manner:

[0025] The HomeCloud smart control center synchronizes the status of all available computing devices, including the available computing power of the computing devices and the transmission network status between different devices, to determine whether to offload the computing;

[0026] The transmission network conditions between devices include the transmission distance between the available computing device and the HomeCloud smart control center and the number of walls that the WIFI signal needs to pass through, the transmission distance between the mobile terminal and the HomeCloud smart control center and the number of walls that the WIFI signal needs to pass through, the transmission distance between the mobile terminal and the available computing device and the number of walls that the WIFI signal needs to pass through, and the communication noise conditions of different lines;

[0027] After determining the transmission network status, it is necessary to calculate the transmission delay and energy consumption between the mobile terminal and different computing devices. There are a lot of walls in the home, and the signal will also produce a certain amount of power loss when passing through the wall. The signal loss is calculated as:

[0028] PL(dB)=20×log 10(f ttran )+N T ×log 10(d)

[0029] +N wall ×L wall -28

[0030] Among them, f tran is the communication frequency band used by WIFI, the unit in the formula is MHz, N T is the path loss coefficient, d is the distance the WIFI signal travels, N wall is the number of walls passed during transmission, L wall is the power loss of the WIFI signal passing through each wall; all data in the unloading process are transmitted via WIFI, assuming that the signal transmission power of the device is p, the channel bandwidth of WIFI transmission is B, and the channel communication noise is σ 2 , g is the signal gain during the transmission process; during the use of WIFI, the setting of the WIFI signal transmission power has been completed, so for each device, its signal transmission power is a fixed value, then its transmission rate R is

[0031]

[0032] For the signal gain g, no additional power compensation is performed during signal transmission; therefore

[0033] g(dB)=-PL(dB)

[0034] For a bit size of D k The time t required for data transmissiontran It is expressed as:

[0035]

[0036] As for the energy consumption of WIFI, the energy consumption is fixed within the WIFI communication time, and its energy consumption E f Calculated as

[0037] E f =pt tran

[0038] Among them, p is the transmission power of WIFI;

[0039] In the computation offloading of a single smart terminal, the transmission time and energy consumption of the smart terminal with and without the assistance of the HomeCloud smart control center will be calculated respectively; the device still needs to perform local calculations to collect user input and decode video streams, so even if it is completely offloaded, a certain amount of calculations needs to be performed locally, and the amount of calculations required for complete local calculations is β offload , called the offloading computation overhead, and its corresponding energy consumption is denoted as E offload ; Then the device needs to offload the amount of calculation C k When the task is, let the computing power of the device be f loc , the unit energy consumption is c, then E offload =β offload cf loc 2 C k ; The local computing amount of the device is C k When the task is to set the device computing power to f loc , the unit energy consumption is c, and its energy consumption is E loc =cf loc 2 C k ;

[0040] If the user adopts the unloading mode to save the terminal battery power, for one of the devices, if there is a transmission line, it may be a line passing through the HomeCloud smart control center or a line not passing through the HomeCloud smart control center, satisfying

[0041] E f +E offload <E loc

[0042] The device is used for computing offloading; if no device meets the above conditions, the smart terminal does not perform computing offloading.

[0043] Furthermore, if the user adopts the offloading mode to improve the computing power of the device, for one of the devices, if there is a transmission line, it may be a line passing through the HomeCloud smart control center or a line not passing through the HomeCloud smart control center, satisfying

[0044] E f +E offload ≤E max

[0045] Where E max If the maximum computing offloading energy consumption accepted by the user is the maximum computing offloading energy consumption accepted by the user, the device is used for computing offloading; if no device meets the above conditions, the intelligent terminal does not perform computing offloading;

[0046] Users can also customize the computing offloading mode, set the maximum computing offloading energy consumption to be accepted when they want to save the battery power of the smart terminal, set computing offloading to be performed when the computing offloading energy consumption is lower than a certain percentage of the local computing energy consumption, and set computing offloading to be performed in any case;

[0047] If there are already smart terminals that are using HomeCloud for computing offload, but the user only wants to optimize the newly added smart terminals separately and does not want to change the offload plans that have been formulated for other terminals, the judgment method is the same as above;

[0048] If there are already smart terminals that are using HomeCloud for computing offloading, and the user wants to formulate a new computing offloading plan for the newly added smart terminal and a specific number of smart terminals or all smart terminals, HomeCloud will synchronize the transmission network status between the newly added smart terminal and all computing devices and the computing capabilities of all computing devices. If there is a computing device that meets the above user-set conditions for the newly added smart terminal, the new smart terminal will join the joint computing offloading;

[0049] Users can also choose to set it to uninstall newly added smart terminals separately. When no device that meets the above requirements is found, global computing uninstallation is performed. Users can also choose to let the HomeCloud smart control center formulate different plans separately in the future, and the user or the HomeCloud smart control center will compare and determine whether to perform global computing uninstallation.

[0050] Furthermore, the determination of the device and method for performing the calculation offloading is specifically performed by the HomeCloud intelligent control center according to the states of all devices suitable for calculation offloading determined in the step of whether to perform the calculation offloading, to determine the device and method for the calculation offloading;

[0051] If the user adopts the offloading mode to save the battery power of the terminal, the device will offload the computing task with the lowest energy consumption possible, and will tend to choose a terminal with lower transmission energy consumption and the corresponding transmission method; since the task has a maximum execution time limit, a single computing device may not be sufficient to complete the task computing offloading;

[0052] For task execution time, in the case of computation offloading, it is mainly divided into the following three parts: local computation time t caused by computation offloading overhead loc , task transmission time t tran , the task computing time on the cloud side t com ; The task transmission time has been calculated in the step of whether to perform calculation offloading. Assuming that the computing power of the terminal device is f loc , the computing power of the cloud device is f com , the computational load of the offloaded task is C k , then the corresponding task processing time is

[0053]

[0054] Consider a scenario of processing a game application. The steps include collecting user input data and sending it, computing in the cloud, transmitting the video stream back to the terminal device, and decoding the video on the terminal device. This series of processes is performed sequentially. Therefore, assuming that the total task execution time is the sum of the three, the total task execution time t off for

[0055] t off =t com +t tran +t loc

[0056] If the computing device with the lowest transmission energy consumption is used and the corresponding transmission mode is adopted for computing offloading, and the device can complete the computing offloading of the task within the maximum execution time, then it is determined that the device performs complete computing offloading in the transmission mode;

[0057] If using the computing device with the lowest transmission energy consumption to perform computation offloading is not sufficient to meet the maximum task execution time requirement, then on the basis of meeting the task execution time requirement, as many computing tasks as possible will be arranged to the device to perform computation offloading, and the computing device with the second lowest transmission energy consumption and the corresponding transmission method will continue to be searched;

[0058] If the computing device with the second lowest transmission energy consumption can execute the remaining tasks within the maximum task execution time, or there is no device suitable for computing offloading, then stop searching and determine the current computing offloading scheme as the execution scheme; otherwise, continue searching for the computing device with the second lowest transmission energy consumption;

[0059] If the user adopts the offloading mode that improves the computing power of the device, then according to the user settings, n devices with the shortest total task execution time are selected, and the tasks are distributed so that the task execution time of each device is the same;

[0060] The user can also choose to comprehensively consider saving terminal battery power and improving device computing power. In this case, the computing device is determined based on the user settings while comprehensively considering energy consumption and task execution time.

[0061] When there are two or more smart terminals in the user's home that need to offload computing, if you choose to formulate a separate computing offloading plan only for the last smart terminal added, the method of formulating the offloading plan is the same as above; you can also choose to re-formulate the computing offloading plan for a specific number or all smart terminals;

[0062] When formulating computing plans for multiple devices, if a computing offloading plan that saves battery power is adopted, consider saving the total battery power of all devices to the greatest extent or saving the battery power of all devices in a certain way; in the mode of saving the total battery power of all devices, the HomeCloud smart control center will obtain the computing power of all available computing devices and the transmission network status between each smart terminal and the computing device, so as to calculate the energy consumption and the amount of tasks that can be executed when tasks on different smart terminals are offloaded to different available computing devices;

[0063] The set of smart terminals that need to jointly specify the offloading scheme is denoted as I. If the available devices meet the judgment conditions in the step of whether to perform computing offloading for any smart terminal in I, then it is a device suitable for providing offloading services. All devices suitable for providing offloading services are denoted as J. For j∈J, its total computing capacity is F j , for the intelligent terminal i∈I, the amount of computing resources allocated is f ij , then the amount of computing resources allocated satisfies

[0064] 0≤f ij ≤F j ,i∈I,j∈J

[0065] ∑ i∈I f ij ≤F j ,j∈J

[0066] The above constraint is denoted as C1; for each terminal device, its task is represented in the form of task flow, then in each time slot of length t, the tasks to be processed are (C i , D i , t) in the form of C i The number of CPU cycles required to complete the computational task, D iis the size of the total transmission task, t is the time slot length; for the terminal device i∈I, let its computing power be f i , whose signal transmission power is p i According to the description of whether to perform computation offloading, the transmission rate from terminal device i to each computing device j∈J is R ij ; For terminal device i∈I, the amount of computational offloading is set to L i , the amount of tasks assigned to each computing device is L ij , the proportion of this part of tasks to all tasks that terminal i needs to complete is α ij ; then the task allocation satisfies

[0067] 0≤L ij ≤L i ,i∈I,j∈J

[0068] 0≤L i ≤C i ,i∈I

[0069] L i =∑ j∈J L ij ,i∈I

[0070] The above constraint is recorded as C2; then for each device, its calculation time satisfies

[0071]

[0072] The above constraint is denoted as C3; for the i-th intelligent terminal i∈I, its local computation includes the tasks left in the local area, the tasks not offloaded, and the energy E of the offloaded computational overhead loc-i Consumption is

[0073] E loc-i =c i f i 2 S i ,i∈I

[0074] S i =L i +β offload (C i -L i )

[0075] c i is the energy consumption coefficient of the i-th intelligent terminal, S i represents the total amount of tasks that need to be processed locally; for the i-th intelligent terminal i∈I, the energy consumption of its data transmission is E tran-i for

[0076]

[0077] For all intelligent terminals, the total energy consumption E sum for

[0078] E sum =∑ i∈I E loc-i +∑ i∈I E tran-i

[0079] Then the HomeCloud smart control center minimizes the function E under the conditions of satisfying the constraints C1, C2, and C3. sum ;

[0080] When developing computing plans for multiple devices, if a computing offload solution is adopted to save battery power, a power saving solution that considers extending the device's battery life can be selected; for device battery life, CPU no-load power consumption and screen display power consumption are introduced; for the total CPU power P cpu It is expressed as:

[0081] P cpu =θcf 3 +P no_load

[0082] Among them, θ is the CPU load ratio, c is the energy consumption coefficient, which is related to factors such as the chip architecture, f is the CPU operating frequency, P no_load The no-load power of the CPU, which accounts for about 20% to 30% of the full-load power consumption of the CPU;

[0083] Since the CPU load is proportional to the amount of computation required for the task it performs, the above formula is converted into a form expressed in terms of the amount of computation required. When the amount of computation required for the task is C and the total running time of the device is T, the total CPU energy consumption E cpu Calculated as

[0084] E cpu =cf 2 C+P no_load T

[0085] For screen display power consumption, use E screen Indicates that it is set as a constant during the operation of the mobile terminal; at the same time, E screen It also represents the static power consumption of all mobile terminals including the screen display power consumption;

[0086] For smart terminals, their energy consumption includes CPU energy consumption E CPU , network transmission energy consumption E tran , screen display power consumption E screen ; Then the total energy consumption E of the intelligent terminal is expressed as

[0087] E=E cpu +Etran +E scree n

[0088] For intelligent terminal i∈I, its total energy consumption is E i , the remaining battery power is B i , then the remaining battery life is t bai-i for

[0089]

[0090] If the user wants to extend the average expected battery life of the device, he needs to maximize the function under the conditions of satisfying the constraints C1, C2, and C3.

[0091] T sum =∑ i∈I t bat-i ,i∈I

[0092] T sum The sum of the battery life of all devices.

[0093] If the user wants to give priority to extending the battery life of low-power devices, he needs to maximize the function under the conditions of satisfying the constraints C1, C2, and C3.

[0094] T min =min({t bat-i |i∈I})

[0095] T min To calculate the battery life of the device that runs out of power the fastest during the uninstall process.

[0096] Users can also choose to optimize individual battery life with a certain weight ratio, or choose to optimize battery life and overall power consumption with a certain weight ratio; users can choose to comprehensively optimize task execution time and device power consumption, all of which must be optimized under the conditions of satisfying restrictions C1, C2, and C3; users can also choose to consider whether to uninstall newly added devices separately after the HomeCloud smart control center has obtained the optimal solution, or choose to let the HomeCloud smart control center decide whether to uninstall them separately.

[0097] Compared with the prior art, the present invention has the following beneficial effects:

[0098] (1) Since HomeCloud computing offloading utilizes existing computing resources in the home, it does not require additional computing resource construction costs. In addition, the data transmission rate of the home LAN is relatively fast and is less affected by the access network, so it can quickly offload computing.

[0099] (2) Since the idle and usage status of all devices in the home are controllable by the user, and when the user is using one device, the probability that other devices are idle is high; the user can stably offload the computing tasks on the mobile terminal to the idle device;

[0100] (3) In HomeCloud computing offloading, since the existing computing resources within the home are used for computing offloading, idle computing resources can be reused, effectively improving the utilization rate of computing resources. At the same time, HomeCloud computing offloading uses devices that have been idle for a long time but still have computing power for auxiliary computing offloading, realizing the reuse of idle resources and reducing the impact of a large amount of electronic waste on the environment.

[0101] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0103] Figure 1 Schematic diagram of existing MEC computing offloading;

[0104] Figure 2 Schematic diagram of existing D2D computing offloading;

[0105] Figure 3 This is a graph showing the relationship between the battery life of a mobile phone and noise when there is no wall in the specific implementation;

[0106] Figure 4 This is a graph showing the relationship between the battery life of a VR headset and noise when there is no wall in the specific implementation;

[0107] Figure 5 This is a graph showing the relationship between the battery life of a mobile phone and noise when there are multiple walls.

[0108] Figure 6 This is a graph showing the relationship between the battery life of a VR headset and noise when there are multiple walls.

[0109] Figure 7 Schematic diagram for computing offload using HomeCloud dedicated devices;

[0110] Figure 8 Schematic diagram for computational offloading using general equipment;

[0111] Fig. 9Flowchart for calculating offloading for the present invention. DETAILED DESCRIPTION

[0112] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0113] Currently common computing offloading implementation methods include multi-access edge computing (MEC) computing offloading methods and device-to-device (D2D) computing offloading methods.

[0114] Among them, Figure 1 As shown in the figure, the computational offloading of multi-access edge computing is carried out through edge servers, such as base stations, which provide computing power support to devices within a certain range through network transmission. The devices that need to perform computational offloading transmit the computing tasks to the edge servers through the network for processing, and the edge servers then return the computing results to the devices through the network, thereby reducing the computing overhead of the devices themselves. Figure 2 As shown, D2D computing offloading is carried out through base station assistance or D2D direct communication, and part of the computing tasks on the device that currently has computing needs are transferred to the currently idle device that is not performing computing tasks through the network for auxiliary computing, thereby effectively utilizing the total computing power of all devices.

[0115] The above prior art has the following disadvantages:

[0116] If computing tasks are processed directly on existing mobile devices, the energy consumption of the device will increase significantly, which will greatly increase the weight of the required battery, reduce the portability of the device, and prevent users from having a good user experience. Among the two commonly used computing offloading processing methods, multi-access edge computing is closer to the user, but because it uses wireless transmission in the access network, the transmission rate is limited, and the data transmission cannot be guaranteed to be stable for a long time, which affects the user experience. And if a large number of computing tasks are placed on the edge server for processing, the workload of the edge server will be significantly increased, the demand for the number and processing power of edge servers will increase, and the construction cost of edge servers will be greatly increased.

[0117] Although D2D computing offloading utilizes idle computing resources and saves computing costs, in D2D computing offloading, different devices may be owned by different users, and their idle and working states may change at any time, which is uncontrollable. Therefore, it is not suitable for processing long-term computing offloading tasks on the same device. And because the working process of some tasks is carried out in the form of workflows and needs to work continuously, repeatedly switching devices will significantly increase the migration cost of tasks, which is not suitable for the execution of continuous tasks.

[0118] In order to solve the above problems, the present invention proposes a home cloud HomeCloud computing offloading architecture, which mainly includes the following four components: a local area network for communication, a mobile smart terminal for offloading, a computing device that provides offloading services, and a HomeCloud smart control center.

[0119] The present invention uses a local area network within a home for data transmission and communication, and can use any local area network communication technology. Since the local area network transmission rate is relatively fast and is less affected by the access network, the computing tasks and computing results can be transmitted more quickly and stably, improving the user experience.

[0120] A mobile smart terminal can be any mobile terminal that needs to perform computing tasks. When it is ready to perform computing tasks, it can send its task information to the resource sharing middleware. If there is a device that can perform computing offload tasks, it can offload the task to the computing device for processing with the assistance of the resource sharing middleware. When the mobile terminal offloads tasks, it can choose to offload all tasks or partially offload them and perform collaborative computing with the HomeCloud computing device. Task offloading can reduce the computing overhead of the mobile terminal and extend its battery life. Collaborative computing with HomeCloud computing devices with strong computing power can also improve the overall execution efficiency of the task and enhance the ability of smart terminals to handle larger computing tasks.

[0121] The computing devices that provide offloading services are divided into two types: existing computing devices and HomeCloud computing offloading dedicated computing devices. Existing computing devices include devices that are still in normal use by users but are idle when the smart terminal makes an offloading request, and devices that are in a long-term idle state but still have computing power. For computing devices that are still in normal use by users, their computing power is relatively strong, and they can complete all tasks of the smart terminal at a faster computing speed. For devices that have been idle for a long time, their computing power may be relatively weak due to old models, but they can still perform collaborative computing to improve the overall computing power. Using existing devices for computing offloading makes full use of the existing computing resources in the user's home and saves the cost of computing resource construction. However, since the existing computing devices themselves need to process tasks, if they generate tasks that need to be processed during the execution of computing offloading, task migration may be required. On the basis of using existing computing devices for computing offloading, this method can also introduce HomeCloud computing offloading dedicated computing devices. HomeCloud computing offloading dedicated computing devices will effectively solve the problem of insufficient overall computing power of home devices. Its computing power is stronger than that of existing devices, and large computing tasks can be quickly executed within the home. In addition, HomeCloud computing offloading dedicated computing equipment is specifically used to serve HomeCloud computing offloading. The equipment itself does not generate tasks that need to be processed and can serve offloading tasks more stably.

[0122] The HomeCloud intelligent control center is responsible for the overall resource scheduling within the HomeCloud computing offloading architecture. It is the core of the HomeCloud computing offloading architecture and is crucial to improving system performance and resource utilization. When the mobile smart terminal generates or changes the task requirements, it first sends the task information to the HomeCloud intelligent control center, including the number of data transmission bits of the task and the number of clock cycles required for the computing task. After receiving the task requirements, the HomeCloud intelligent control center synchronizes the idle status of the computing device and the CPU computing power, and determines the unloading device based on the task requirements and computing resource utilization. In the process of determining the unloading device, this patent can adopt an unloading mode that maximizes battery life, and a management mode that prioritizes the battery life of low-power devices in the case of multiple devices. If the task requirements and the idle state of the device remain unchanged, the computing offloading task will continue to be unloaded according to the unloading plan determined above. If the idle status of the computing device changes, the computing device will send information to the HomeCloud intelligent control center, and the resource sharing middleware will reallocate the unloading plan.

[0123] The main basic units of the HomeCloud smart control center include the signal sending and receiving unit, the storage unit and the computing unit.

[0124] The signal sending and receiving unit is responsible for information transmission with the smart terminal and the HomeCloud computing offloading device. When the mobile smart terminal generates or changes the task requirements, the HomeCloud smart control center receives information from the mobile smart terminal through the signal receiving unit. Secondly, the HomeCloud smart control center sends the idle status of the synchronous computing device and the request for CPU computing power through the signal sending unit, and receives the information through the signal receiving unit. After determining the offloading plan, the HomeCloud smart control center sends the offloading plan to the corresponding computing device and smart terminal through the signal sending unit. If the smart terminal needs to forward the task through the HomeCloud smart control center, the HomeCloud smart control center receives the task through the signal receiving unit, sends the task to the corresponding computing offloading device through the signal sending unit, and receives the task return result through the signal receiving unit, and then sends it to the mobile smart terminal through the signal sending unit.

[0125] The signal sending and receiving unit may be a WIFI wireless communication module, which sends and receives information through WIFI signals. The signal sending and receiving unit may be a wired WIFI interface, which is connected to the WIFI AP through a cable or optical fiber. The signal sending and receiving unit may be other signal sending and receiving modules.

[0126] The storage unit is responsible for storing computing offloading plans, offloading device information, and intermediate data. When the HomeCloud computing offloading device determines the computing offloading plan, it stores the plan in the storage unit. The stored offloading device information includes the idle status of the computing device and the CPU computing power, as well as the corresponding latest IP address, which is updated each time before a computing offloading plan is required. The storage unit can store intermediate data for formulating computing offloading plans, can store transfer files for task transfer, and can process intermediate data generated by other processes. The storage unit can be a separate memory that can be connected to a computing offloading device for storage.

[0127] The computing unit is responsible for formulating a computing offloading plan and making a decision based on whether to determine the offloading step and the device and method steps for performing the computing offloading. The computing unit can be an independent CPU computing unit. The computing unit can also be connected to a computing offloading device, and the computing offloading device formulates the computing offloading plan based on whether to determine the offloading step and the device and method steps for performing the computing offloading. The computing unit can also be an intelligent terminal, and the computing offloading plan is formulated directly on the intelligent terminal.

[0128] In order to study the impact of HomeCloud computing offloading on the battery life of mobile smart terminals, this study simulated two mobile terminals, a smartphone and a VR headset, when playing games. The VR headset represents a mobile smart terminal with large computing and transmission volumes, while the mobile phone represents a mobile smart terminal with small computing and transmission volumes. When the device performs computing offloading, it can save the computing energy consumption of the corresponding offloaded tasks, but it requires additional communication energy consumption. The simulation uses WIFI technology for LAN communication and PC for computing offloading.

[0129] When the WIFI AP, mobile terminal and computing offloading device (computer) are all in a 10m*10m room, there is no signal attenuation caused by the wall. The simulation results are as follows:

[0130] like Figure 3 As shown in the figure, when the computing tasks on the mobile phone are offloaded to a computer with a computing power of 5GHz, the impact on the battery life of the mobile phone under different environmental noises. The bottom blue line is the battery life of the mobile phone when no computing offloading is performed, the top purple line is the battery life of the mobile phone when the channel bandwidth of the offload transmission is 20MHz, the orange line is the battery life of the mobile phone when the channel bandwidth of the offload transmission is 10MHz, and the yellow line is the battery life of the mobile phone when the channel bandwidth of the offload transmission is 5MHz;

[0131] like Figure 4 As shown in the figure, when the computing tasks on the VR headset are offloaded to a computer with a computing power of 5GHz, the impact on the battery life of the VR headset under different environmental noises. The bottom blue line is the battery life of the VR headset when no computing offload is performed, the top purple line is the battery life of the VR headset when the channel bandwidth of the offload transmission is 20MHz, the orange line is the battery life of the VR headset when the channel bandwidth of the offload transmission is 10MHz, and the yellow line is the battery life of the VR headset when the channel bandwidth of the offload transmission is 5MHz;

[0132] In a room with normal noise levels of -100dBm to -80dBm, computing offload via HomeCloud can extend the battery life of a mobile phone by 70% to 100% and the battery life of a VR headset by 80% to 130%.

[0133] When the mobile terminal, computer, and WIFI AP are randomly distributed in different rooms, that is, when there is wall attenuation, the simulation results are as follows:

[0134] like Figure 5As shown in the figure, when the computing tasks on the mobile phone are offloaded to a computer with a computing power of 5GHz, the impact on the battery life of the mobile phone under different environmental noises. The bottom blue line is the battery life of the mobile phone when no computing offloading is performed, the top purple line is the battery life of the mobile phone when the channel bandwidth of the offload transmission is 20MHz, the orange line is the battery life of the mobile phone when the channel bandwidth of the offload transmission is 10MHz, and the yellow line is the battery life of the mobile phone when the channel bandwidth of the offload transmission is 5MHz;

[0135] like Figure 6 As shown in the figure, when the computing tasks on the VR headset are offloaded to a computer with a computing power of 5GHz, the impact on the battery life of the VR headset under different environmental noises. The bottom blue line is the battery life of the VR headset when no computing offload is performed, the top purple line is the battery life of the VR headset when the channel bandwidth of the offload transmission is 20MHz, the orange line is the battery life of the VR headset when the channel bandwidth of the offload transmission is 10MHz, and the yellow line is the battery life of the VR headset when the channel bandwidth of the offload transmission is 5MHz;

[0136] When mobile terminals, computers, and WIFI APs are randomly distributed in different rooms, it can also extend the battery life of mobile phones by 60% to 100% and the battery life of VR headsets by 70% to 130%. This method effectively improves the overall battery life of smart terminals.

[0137] like Figure 7-9 As shown:

[0138] When a user uses a VR headset for game applications, if a dedicated computing offloading device is used for computing offloading, the VR headset will first send its offloading requirements to the HomeCloud smart control center. After receiving the task requirements, the HomeCloud smart control center synchronizes the CPU operation status of all computing devices, determines the dedicated computing offloading device to perform the offloading task based on whether to perform computing offloading and the method of determining the computing offloading device, and sends the information of the executed computing offloading plan and the computing offloading task to the corresponding dedicated computing offloading device and the VR headset. The VR headset can communicate directly with the computing device, or it can perform auxiliary communication through the HomeCloud smart control center, send its task requirements to the computing device, and the computing device will return the result to the VR headset after the calculation.

[0139] When a user uses a VR headset for game applications, if an existing device is used for computing offloading. The VR headset first sends its offloading requirements to the HomeCloud smart control center. After receiving the task requirements, the HomeCloud smart control center synchronizes the CPU operation status of all computing devices, determines the existing computing devices that perform the offloading tasks based on whether to perform computing offloading and the method of determining the computing offloading device, and sends the information of the executed computing offloading plan and the computing offloading task to the corresponding existing computing devices and VR headsets. When the idle status of the existing computing device itself has not changed, the task will always be performed according to the formulated computing offloading plan. If the existing computing device responsible for the computing offloading task itself needs to perform the task, it sends this information back to the HomeCloud smart control center, which will re-determine the device that performs the computing offloading task. The VR headset can communicate directly with the computing device, or it can perform auxiliary communication through the HomeCloud smart control center.

[0140] Whether to confirm the uninstallation step: The HomeCloud smart control center synchronizes the status of all available computing devices, including the available computing power of the computing devices and the transmission network status between different devices, to determine whether to perform computing offload.

[0141] The transmission network conditions between devices include the transmission distance between the available computing device and the HomeCloud smart control center and the number of walls that the WIFI signal needs to pass through for transmission, the transmission distance between the mobile terminal and the HomeCloud smart control center and the number of walls that the WIFI signal needs to pass through for transmission, the transmission distance between the mobile terminal and the available computing device and the number of walls that the WIFI signal needs to pass through for transmission, and the communication noise conditions of different lines.

[0142] After determining the transmission network status, it is necessary to calculate the transmission delay and energy consumption between the mobile terminal and different computing devices. In common computing offloading research, in terms of signal transmission, usually only the distance between the user device and the computing device and a small amount of white noise and inter-channel interference are considered, which has good applicability in outdoor transmission. However, in a home environment, the situation is quite different. There are a lot of walls in the home, and the signal will also produce a certain amount of power loss when passing through the wall. The impact of this part of the loss is quite large. Therefore, we refer to the WIFI loss model and calculate its signal loss as

[0143] PL(dB)=20×log 10(f tran )+N T ×log 10(d)

[0144] +N wall ×L wall -28

[0145] Among them, f tran is the communication frequency band used by WIFI, the unit in the formula is MHz, N T is the path loss coefficient, d is the distance the WIFI signal travels, N wall is the number of walls passed during transmission, L wall The power loss of the WiFi signal passing through each wall.

[0146] All data in the offloading process is transmitted via WIFI. We assume that the signal transmission power of the device is p, the channel bandwidth of WIFI transmission is B, and the channel communication noise is σ 2 , g is the signal gain during the transmission process. During the use of WIFI, we set the WIFI signal transmission power to be set, so for each device, its signal transmission power is a fixed value, and its transmission rate R is

[0147]

[0148] For the signal gain g, we do not perform additional power compensation during signal transmission.

[0149] g(dB)=-PL(dB)

[0150] Therefore, for a bit size of D k The time t required for data transmission tran It can be expressed as:

[0151]

[0152] Regarding the energy consumption of WIFI, this patent refers to the general computation offloading research, and its energy consumption is fixed within the WIFI communication time, and its energy consumption E f Calculated as

[0153] E f =pt tran

[0154] Where p is the transmission power of WIFI.

[0155] In the computational offloading of a single smart terminal, we will calculate the transmission time and energy consumption of the smart terminal with and without the assistance of the HomeCloud smart control center.

[0156] In terms of computational offloading, the device can perform local computation, completely offload tasks to the cloud, and partially offload tasks to the cloud. However, in practice, offloading is likely to be incomplete. Consider a game application that is executed entirely in the cloud and only returns the results in the form of a video stream. In this case, the device still needs to perform local computations to collect user input and decode the video stream. Therefore, even if it is completely offloaded, this patent believes that a certain amount of computation needs to be performed locally, and the amount of computation is β of the amount of computation required for complete local computation. offload , which is referred to as offloading computation overhead in this patent, and its corresponding energy consumption is recorded as E offload . Then the amount of computation that needs to be offloaded by the device is C k When the task is, let the computing power of the device be f loc , the unit energy consumption is c, then

[0157] E offload =β offload cf loc 2 C k

[0158] The local computing amount of the device is C k When the task is, let the computing power of the device be f loc , the unit energy consumption is c, and its energy consumption is E loc for

[0159] E loc =cf loc 2 C k

[0160] If the user adopts the unloading mode to save the terminal battery power, for one of the devices, if there is a transmission line, it may be a line passing through the HomeCloud smart control center or a line not passing through the HomeCloud smart control center, satisfying

[0161] E f +E offload <E loc

[0162] If no device meets the above conditions, the intelligent terminal will not perform calculation offloading.

[0163] If the user adopts the offload mode to improve the computing power of the device, then for one of the devices, if there is a transmission line, it may be a line passing through the HomeCloud smart control center or a line not passing through the HomeCloud smart control center, satisfying

[0164] E f +E offload ≤Emax

[0165] Where E max If the maximum computing offloading energy consumption that the user can accept is 0, the device can be used for computing offloading. If no device meets the above conditions, the smart terminal will not perform computing offloading.

[0166] Users can also customize the computing offloading mode. They can set the maximum acceptable computing offloading energy consumption when they want to save battery power on smart terminals. They can set computing offloading to be performed when the computing offloading energy consumption is lower than a certain percentage of the local computing energy consumption. They can set computing offloading to be performed in any situation.

[0167] If there are already smart terminals that are using HomeCloud for computing offload, but the user only wants to optimize the newly added smart terminals separately and does not want to change the offload plans that have been formulated for other terminals, the judgment method is the same as above.

[0168] If there are already smart terminals that are using HomeCloud for computing offloading, and the user wants to develop a new computing offloading plan for the newly added smart terminal and a specific number of smart terminals or all smart terminals, HomeCloud will synchronize the transmission network status between the newly added smart terminal and all computing devices and the computing power of all computing devices. If there is a computing device that meets the above user-set conditions for the newly added smart terminal, the new smart terminal will join the joint computing offloading.

[0169] Users can also choose to set it to uninstall newly added smart terminals separately. When no device that meets the above requirements is found, global computing uninstallation is performed. Users can also choose to let the HomeCloud smart control center formulate different plans separately in the future, and the user or the HomeCloud smart control center will compare and determine whether to perform global computing uninstallation.

[0170] Determine the device and method for performing computing offloading: The HomeCloud intelligent control center determines the computing offloading device and method based on whether the status of all devices suitable for computing offloading determined in the offloading step is determined.

[0171] If the user uses the offloading mode to save the terminal battery power, the device will offload the computing task as low as possible, and will tend to choose a terminal with low transmission energy consumption and the corresponding transmission method. Since the task has a maximum execution time limit, a single computing device may not be enough to complete the task computing offloading.

[0172] For task execution time, in the case of computation offloading, it is mainly divided into the following three parts: local computation time t caused by computation offloading overhead loc, task transmission time t tran , the task computing time on the cloud side t com The task transmission time is determined by the offloading step, and the computing power of the terminal device is assumed to be f loc , the computing power of the cloud device is f com , the computational load of the offloaded task is C k , then the corresponding task processing time is

[0173]

[0174] Consider a scenario of processing a game application, the steps include collecting user input data and sending it, computing in the cloud, transmitting the video stream back to the terminal device, and decoding the video on the terminal device. This series of processes is performed sequentially. Therefore, this patent assumes that the total task execution time is the sum of the three, and the total task execution time t off for

[0175] t off =t com +t tran +t loc

[0176] If a computing device with the lowest transmission energy consumption is used and a corresponding transmission method is adopted for computing offloading, and the device can complete the computing offloading of the task within the maximum execution time, then it is determined that the device performs complete computing offloading using this transmission method.

[0177] If using the device with the lowest transmission energy consumption to perform computational offloading is not sufficient to meet the maximum task execution time requirement, then as many computing tasks as possible will be scheduled to the device to perform computational offloading on the basis of meeting the task execution time requirement, and continue to search for the computing device with the second lowest transmission energy consumption and the corresponding transmission method.

[0178] If the computing device with the second lowest transmission energy consumption can execute the remaining tasks within the maximum task execution time, or there is no device suitable for computing offloading, the search is stopped and the current computing offloading scheme is determined to be the execution scheme. Otherwise, the search for the computing device with the second lowest transmission energy consumption continues.

[0179] If the user adopts the offloading mode to improve the computing power of the device, n devices with the shortest total task execution time are selected according to the user settings, and the tasks are distributed so that the task execution time of each device is the same.

[0180] The user can also choose to comprehensively consider saving terminal battery power and improving device computing power. In this case, the computing device is determined based on the user settings while comprehensively considering energy consumption and task execution time.

[0181] When there are two or more smart terminals in the user's home that need to offload computing, you can choose to formulate a separate computing offloading plan only for the last smart terminal added, and the method of formulating the offloading plan is the same as above. You can also choose to re-formulate the computing offloading plan for a specific number or all smart terminals.

[0182] When developing computing plans for multiple devices, if a computing offloading plan that saves battery power is adopted, you can consider saving the total battery power of all devices to the greatest extent or saving the battery power of all devices in a certain way.

[0183] In the mode of saving the total battery power of all devices, the HomeCloud smart control center will obtain the computing power of all available computing devices and the transmission network status between each smart terminal and the computing device, so as to calculate the energy consumption and the amount of executable tasks when offloading tasks from different smart terminals to different available computing devices.

[0184] The set of smart terminals that need to jointly specify the uninstallation plan is denoted as I. If the available device meets the judgment condition in the uninstallation step for any smart terminal in I, it is a device suitable for providing uninstallation services. All devices suitable for providing uninstallation services are denoted as J.

[0185] For j∈J, its total computing power is F j , for the intelligent terminal i∈I, the amount of computing resources allocated is f ij , then the amount of computing resources allocated satisfies

[0186] 0≤f ij ≤F j ,i∈I,j∈J

[0187] ∑ i∈I f ij ≤F j ,j∈J

[0188] The above restriction condition is denoted as C1.

[0189] For each terminal device, its tasks are represented in the form of task flow. Then, in each time slot of length t, the tasks to be processed are (C i , D i , t) in the form of C i The number of CPU cycles required to complete the computational task, D i is the size of the total transmission task, t is the time slot length. For the terminal device i∈I, let its computing power be f i , whose signal transmission power is p i According to whether the offloading step is determined, the transmission rate from the terminal device i to each computing device j∈J can be deduced as Rij For terminal device i∈I, the amount of computational offloading is set to L i , the amount of tasks assigned to each computing device is L ij , the proportion of this part of tasks to all tasks that terminal i needs to complete is α ij .

[0190] Then the task allocation satisfies

[0191] 0≤L ij ≤L i ,i∈I,j∈J

[0192] 0≤L i ≤C i ,i∈I

[0193] L i =∑ j∈J L ij ,i∈I

[0194] The above constraint is recorded as C2. Then for each device, its calculation time satisfies

[0195]

[0196] The above constraint is denoted as C3. For the i-th intelligent terminal i∈I, the energy of its local computing is E loc-i The consumption (including the tasks left in the local machine and the computational overhead of the unloaded tasks) is

[0197] E loc-i =c i f i 2 S i ,i∈I

[0198] S i =L i +β offload (C i -L i )

[0199] c i is the energy consumption coefficient of the i-th intelligent terminal, S i Represents the total amount of tasks that need to be processed locally.

[0200] For the i-th intelligent terminal i∈I, the energy consumption of its data transmission is E tran-i for

[0201]

[0202] For all intelligent terminals, the total energy consumption E sum for

[0203] E sum =∑ i∈I E loc-i +∑ i∈I E tran-i

[0204] Then the HomeCloud smart control center minimizes the function E under the conditions of satisfying the constraints C1, C2, and C3. sum .

[0205] When developing computing plans for multiple devices, if you adopt a computing offload solution that saves battery power, you can choose a power saving solution that considers extending the battery life of the device.

[0206] For device battery life, common computational offloading research only considers the energy consumption of processing unit tasks and transmitting unit data. However, during the use of the device, even if a large number of computational tasks do not need to be processed, the CPU no-load power consumption and static power consumption such as screen display of the device are still huge. Therefore, this patent introduces CPU no-load power consumption and screen display power consumption.

[0207] For the total power P of the CPU cpu It can be expressed as:

[0208] P cpu =θcf 3 +P no_load

[0209] Among them, θ is the CPU load ratio, c is the energy consumption coefficient, which is related to factors such as the chip architecture, f is the CPU operating frequency, P no_load It is the CPU's no-load power, which accounts for about 20% to 30% of the CPU's full-load power consumption.

[0210] Since the CPU load is proportional to the amount of computation required to perform the task, the above formula can be converted into a form expressed in terms of the amount of computation required to perform the task. When the amount of computation required to perform the task is C and the total running time of the device is T, the total CPU energy consumption E cpu Calculated as

[0211] E cpu =cf 2 C+P no_load T

[0212] For screen display power consumption, we use E screen Indicates that it is set as a constant during the operation of the mobile terminal. screen It can also represent the static power consumption of all mobile terminals including screen display power consumption.

[0213] For smart terminals, their energy consumption includes CPU energy consumption E cpu , network transmission energy consumption E tran , screen display power consumption E screen The total energy consumption E of the smart terminal is expressed as

[0214] E=E cpu +E tran +E screen

[0215] For intelligent terminal i∈I, its total energy consumption is E i , the remaining battery power is B i , then the remaining battery life is t bai-i for

[0216]

[0217] If the user wants to extend the average expected battery life of the device, he needs to maximize the function under the conditions of satisfying the constraints C1, C2, and C3.

[0218] T sum =∑ i∈I t bat-i ,i∈I

[0219] T sum The sum of the battery life of all devices.

[0220] If the user wants to give priority to extending the battery life of low-power devices, he needs to maximize the function under the conditions of satisfying the constraints C1, C2, and C3.

[0221] T min =min({t bat-i |i∈I})

[0222] T min To calculate the battery life of the device that runs out of power the fastest during the uninstall process.

[0223] Users can also choose to optimize individual battery life with a certain weight ratio, or choose to optimize battery life and overall power consumption with a certain weight ratio. Users can also choose to comprehensively optimize task execution time and device power consumption, all of which must be optimized under the conditions of satisfying restrictions C1, C2, and C3. Users can also choose to consider whether to uninstall newly added devices separately after the HomeCloud smart control center has obtained the optimal solution, and the HomeCloud smart control center can also decide whether to uninstall them separately.

[0224] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. The computing offloading architecture based on HomeCloud is characterized by: It includes a local area network for communication, a mobile smart terminal for offloading, a computing device that provides offloading services, and a HomeCloud smart control center; The local area network uses a local area network within the home for data transmission and communication; When the mobile intelligent terminal is preparing to perform a computing task, it sends the task information to the resource sharing middleware. If there is a device that can perform the computing offload task, the task is offloaded to the computing device for processing via the resource sharing middleware. The computing devices include existing computing devices and computing devices dedicated to HomeCloud computing offloading; the existing computing devices include devices that are still in normal use by users but are idle when the smart terminal makes an offloading request, and also include devices that are idle for a long time but still have computing power; the computing power of the computing devices dedicated to HomeCloud computing offloading is stronger than that of existing computing devices, and can quickly execute large computing tasks in the home; The HomeCloud intelligent control center is responsible for overall resource scheduling within the HomeCloud computing offloading architecture. When the mobile intelligent terminal generates or changes task requirements, it first sends task information to the HomeCloud intelligent control center, including the number of data transmission bits of the task and the number of clock cycles required for the computing task. After receiving the task requirements, the HomeCloud intelligent control center synchronizes the idle status of the computing device and the CPU computing power, and determines the offloading device based on the task requirements and computing resource utilization. The HomeCloud intelligent control center includes a signal sending and receiving unit, a storage unit and a computing unit; The signal sending and receiving unit is used to transmit information with the smart terminal and the HomeCloud computing offloading device; when the mobile smart terminal generates or changes the task requirement, the HomeCloud smart control center receives the information from the mobile smart terminal through the signal receiving unit; the HomeCloud smart control center sends the idle status of the synchronous computing device and the request for CPU computing power through the signal sending unit, and receives the information through the signal receiving unit; after determining the offloading plan, the HomeCloud smart control center sends the offloading plan to the corresponding computing device and the smart terminal through the signal sending unit; if the smart terminal needs to forward the task through the HomeCloud smart control center, after the HomeCloud smart control center receives the task through the signal receiving unit, it sends the task to the corresponding computing offloading device through the signal sending unit; and after receiving the task return result through the signal receiving unit, it sends it to the mobile smart terminal through the signal sending unit; The storage unit is used to store computing offloading solutions, offloading device information, and intermediate data; when the HomeCloud computing offloading device determines the computing offloading solution, the solution is stored in the storage unit; the stored offloading device information includes the idle status of the computing device and the CPU computing power, as well as the corresponding latest IP address, which is updated each time before the computing offloading solution needs to be formulated; The storage unit stores intermediate data for formulating a computing offloading plan, stores transfer files for task transfer, and processes intermediate data generated by other processes; The storage unit is a separate memory connected to a computation offloading device for storage; The computing unit formulates a computing offloading plan and makes a decision based on whether to perform computing offloading and the steps of determining the device and method for performing computing offloading.

2. The computing offloading architecture based on HomeCloud according to claim 1 is characterized in that: When performing task offloading, the mobile intelligent terminal chooses to offload all or part of the task, and performs collaborative computing with the HomeCloud computing device.

3. The computing offloading architecture based on HomeCloud according to claim 1 is characterized in that: If the task demand and the idle state of the device remain unchanged, the computing offloading task will continue to be offloaded according to the offloading plan determined above; if the idle state of the computing device changes, the computing device will send information to the HomeCloud smart control center, and the resource sharing middleware will reallocate the offloading plan.

4. The computing offloading architecture based on HomeCloud according to claim 1 is characterized in that: The signal sending and receiving unit adopts a WIFI wireless communication module to send and receive information through WIFI signals; the signal sending and receiving unit adopts a wired WIFI interface to be connected to the WIFIAP through a cable or optical fiber.

5. The computing offloading architecture based on HomeCloud according to claim 1 is characterized in that: The computing unit is an independent CPU computing unit or is connected to a computing offloading device, and the computing offloading device formulates a computing offloading plan.

6. The computing offloading architecture based on HomeCloud according to claim 1 is characterized in that: The computing unit is an intelligent terminal, and the computing offloading plan is formulated directly on the intelligent terminal.

7. An unloading method based on a computing unloading architecture of a home cloud HomeCloud, wherein the unloading is performed using the computing unloading architecture based on a home cloud HomeCloud as claimed in any one of claims 1 to 6, characterized in that: The steps include: S1. The mobile smart terminal sends its uninstallation request to the HomeCloud smart control center; S2. After receiving the task requirement, the HomeCloud intelligent control center synchronizes the CPU operation status of all computing devices, determines the computing device to perform the offloading task according to whether to perform the computing offloading and the device and method for performing the computing offloading, and sends the information of the computing offloading plan and the computing offloading task to the computing device and mobile intelligent terminal corresponding to the computing offloading; S3. The mobile smart terminal communicates directly with the computing device, or performs auxiliary communication through the HomeCloud smart control center, sends the task requirements to the computing device, and the computing device performs calculations and then returns the results to the mobile smart terminal.

8. The unloading method based on the computing unloading architecture of the home cloud HomeCloud according to claim 7 is characterized in that: Whether to perform calculation offloading is specifically determined in the following manner: The HomeCloud smart control center synchronizes the status of all available computing devices, including the available computing power of the computing devices and the transmission network status between different devices, to determine whether to offload the computing; The transmission network conditions between devices include the transmission distance between the available computing device and the HomeCloud smart control center and the number of walls that the WIFI signal needs to pass through, the transmission distance between the mobile terminal and the HomeCloud smart control center and the number of walls that the WIFI signal needs to pass through, the transmission distance between the mobile terminal and the available computing device and the number of walls that the WIFI signal needs to pass through, and the communication noise conditions of different lines; After determining the transmission network status, it is necessary to calculate the transmission delay and energy consumption between the mobile terminal and different computing devices. There are a lot of walls in the home, and the signal will also produce a certain amount of power loss when passing through the wall. The signal loss is calculated as: PL(dB)=20×log10(f tran )+N T ×log 10(d)+N wall ×L wall -28 Among them, f tran is the communication frequency band used by WIFI, the unit in the formula is MHz, N T is the path loss coefficient, d is the distance the WIFI signal travels, N wall is the number of walls passed during transmission, L wall is the power loss of the WIFI signal passing through each wall; all data in the unloading process are transmitted via WIFI, assuming that the signal transmission power of the device is p, the channel bandwidth of WIFI transmission is B, and the channel communication noise is σ 2 , g is the signal gain during the transmission process; during the use of WIFI, the WIFI signal transmission power has been set, so for each device, its signal transmission power is a fixed value, and its transmission rate R is For the signal gain g, no additional power compensation is performed during signal transmission; therefore g(dB)=-PL(dB) For a bit size of D k The time t required for data transmission tran It is expressed as: As for the energy consumption of WIFI, the energy consumption is fixed within the WIFI communication time, and its energy consumption E f Calculated as E f =pt tran Where p is the transmission power of WIFI; In the computation offloading of a single smart terminal, the transmission time and energy consumption of the smart terminal with and without the assistance of the HomeCloud smart control center will be calculated respectively; the device still needs to perform local computations to collect user input and decode video streams, so even if it is completely offloaded, a certain amount of computation needs to be performed locally, and the amount of computation is β of the amount of computation required for complete local computation. offload , called the offloading computation overhead, and its corresponding energy consumption is denoted as E offload ; Then the amount of computation that needs to be offloaded by the device is C k When the task is, let the computing power of the device be f loc , the unit energy consumption is c, then E offload =β offload cf loc 2 C k ; The local computing amount of the device is C k When performing tasks, set the device computing capability to f loc , the unit energy consumption is c, and its energy consumption is E loc =cf loc 2 C k ; If the user adopts the unloading mode to save the terminal battery power, for one of the devices, if there is a transmission line, it may be a line passing through the HomeCloud smart control center or a line not passing through the HomeCloud smart control center, satisfying AND f +E offload <And loc The device is used for computing offloading; if no device meets the above conditions, the smart terminal does not perform computing offloading.

9. The unloading method based on the computing unloading architecture of the home cloud HomeCloud according to claim 8, characterized in that: If the user adopts the offload mode to improve the computing power of the device, then for one of the devices, if there is a transmission line, it may be a line passing through the HomeCloud smart control center or a line not passing through the HomeCloud smart control center, satisfying AND f +E offload ≤E max Where E max If the maximum computing offloading energy consumption accepted by the user is the maximum computing offloading energy consumption accepted by the user, the device is used for computing offloading; if no device meets the above conditions, the intelligent terminal does not perform computing offloading; Users can also choose a custom computing offloading mode, set the maximum computing offloading energy consumption to be accepted when they want to save battery power on smart terminals, set computing offloading to be performed when computing offloading energy consumption is lower than a certain percentage of local computing energy consumption, and set computing offloading to be performed in any case; If there are already smart terminals that are using HomeCloud for computing offload, but the user only wants to optimize the newly added smart terminals separately and does not want to change the offload plans that have been formulated for other terminals, the judgment method is the same as above; If there are already smart terminals that are using HomeCloud for computing offloading, and the user wants to formulate a new computing offloading plan for the newly added smart terminal and a specific number of smart terminals or all smart terminals, HomeCloud will synchronize the transmission network status between the newly added smart terminal and all computing devices and the computing capabilities of all computing devices. If there is a computing device that meets the above user-set conditions for the newly added smart terminal, the new smart terminal will join the joint computing offloading; Users can also choose to set up a separate uninstallation for newly added smart terminals. When no device that meets the above requirements can be found, global computing offloading can be performed. Users can also choose to have the HomeCloud smart control center formulate different plans in the future, and then the user or the HomeCloud smart control center will compare and determine whether to perform global computing offloading.

10. The unloading method based on the computing unloading architecture of the home cloud HomeCloud according to claim 7, characterized in that: The determination of the device and method for performing the calculation offloading is specifically performed by the HomeCloud intelligent control center according to the states of all devices suitable for calculation offloading determined in the step of whether to perform the calculation offloading, and determining the device and method for the calculation offloading; If the user adopts the offloading mode to save the battery power of the terminal, the device will offload the computing task with the lowest energy consumption possible, and will tend to choose a terminal with lower transmission energy consumption and the corresponding transmission method; since the task has a maximum execution time limit, a single computing device may not be sufficient to complete the task computing offloading; For task execution time, in the case of computation offloading, it is mainly divided into the following three parts: local computation time t caused by computation offloading overhead loc , task transmission time t tran , the task computing time in the cloud t com ; The task transmission time has been calculated in the step of whether to perform calculation offloading. Assuming that the computing power of the terminal device is f loc , the computing power of the cloud device is f com , the computational load of the offloaded task is C k , then the corresponding task processing time is Consider a scenario of processing a game application. The steps include collecting user input data and sending it, computing in the cloud, transmitting the video stream back to the terminal device, and decoding the video on the terminal device. This series of processes is performed sequentially. Therefore, assuming that the total task execution time is the sum of the three, the total task execution time t off t off =t com +t tran +t loc If the computing device with the lowest transmission energy consumption is used and the corresponding transmission mode is adopted for computing offloading, and the device can complete the computing offloading of the task within the maximum execution time, then it is determined that the device performs complete computing offloading in the transmission mode; If using the device with the lowest transmission energy consumption to offload calculations is not sufficient to meet the maximum task execution time requirement, then on the basis of meeting the task execution time requirement, as many calculation tasks as possible will be arranged to the device to perform calculation offloading, and continue to search for the calculation device with the second lowest transmission energy consumption and the corresponding transmission method; If the computing device with the second lowest transmission energy consumption can execute the remaining tasks within the maximum task execution time, or there is no device suitable for computing offloading, then stop searching and determine the current computing offloading scheme as the execution scheme; otherwise, continue searching for the computing device with the second lowest transmission energy consumption; If the user adopts the offloading mode that improves the computing power of the device, then according to the user settings, n devices with the shortest total task execution time are selected, and the tasks are distributed so that the task execution time of each device is the same; The user can also choose to comprehensively consider saving terminal battery power and improving device computing power. The computing device is determined based on the user's settings, taking into account energy consumption and task execution time. When there are two or more smart terminals in the user's home that need to offload computing, if you choose to formulate a separate computing offloading plan only for the last smart terminal added, the method of formulating the offloading plan is the same as above; you can also choose to re-formulate the computing offloading plan for a specific number or all smart terminals; When formulating computing plans for multiple devices, if a computing offloading plan that saves battery power is adopted, consider saving the total battery power of all devices to the greatest extent or saving the battery power of all devices in a certain way; in the mode of saving the total battery power of all devices, the HomeCloud smart control center will obtain the computing power of all available computing devices and the transmission network status between each smart terminal and the computing device, so as to calculate the energy consumption and the amount of tasks that can be executed when tasks on different smart terminals are offloaded to different available computing devices; The set of smart terminals that need to jointly specify the offloading solution is recorded as I. If the available device meets the judgment condition in the step of whether to perform the calculation offloading for any smart terminal in I, it is a device suitable for providing offloading services. All devices suitable for providing offloading services are recorded as J. For j∈J, its total computing power is F j , for the intelligent terminal i∈I, the amount of computing resources allocated is f ij , then the amount of computing resources allocated satisfies 0≤f ij ≤F j ,i∈I,j∈J ∑ i∈I f ij ≤F j ,j∈J The above constraint is denoted as C1; for each terminal device, its task is represented in the form of a task flow. Then, in each time slot of length t, the tasks to be processed are (C i , D i , t) in the form of C i The number of CPU cycles required to complete the computational task, D i is the size of the total transmission task, t is the time slot length; for the terminal device i∈I, let its computing power be f i , whose signal transmission power is p i According to the description of whether to perform computation offloading, the transmission rate from terminal device i to each computing device j∈J is R ij ; For terminal device i∈I, the amount of computational offloading is set to L i , the amount of tasks assigned to each computing device is L ij , the proportion of this part of tasks to all tasks that terminal i needs to complete is α ij ; then the task allocation satisfies 0≤L ij ≤L i ,i∈I,j∈J 0≤L i ≤C i ,i∈I L i =∑ j∈J L ij ,i∈I The above constraint is recorded as C2; then for each device, its calculation time satisfies The above restriction condition is recorded as C3; For the i-th intelligent terminal i∈I, its local computation includes the tasks left in the local area, the tasks not offloaded, and the energy E of the offloaded computational overhead loc-i Consumption is E loc-i =c i f i 2 S i ,i∈I S i =L i +b offload (C i -L i ) c i is the energy consumption coefficient of the i-th intelligent terminal, S i represents the total amount of tasks that need to be processed locally; for the i-th intelligent terminal i∈I, the energy consumption of its data transmission is E tran-i for For all intelligent terminals, the total energy consumption E sum for AND sum =∑ i∈I AND loc-i +∑ i∈I AND tran-i Then the HomeCloud smart control center minimizes the function E under the conditions of satisfying the constraints C1, C2, and C3. sum ; When developing computing plans for multiple devices, if a computing offload solution that saves battery power is adopted, a power saving solution that considers extending the device's battery life can be selected; for device battery life, CPU no-load power consumption and screen display power consumption are introduced; the total CPU power P cpu It is expressed as: P cpu =θcf 3 +P no-load Among them, θ is the CPU load ratio, c is the energy consumption coefficient, which is related to factors such as the chip architecture, f is the CPU operating frequency, P no_load The no-load power of the CPU, which accounts for about 20% to 30% of the full-load power consumption of the CPU; Since the CPU load is proportional to the amount of computation required for the task it performs, the above formula is converted into a form expressed in terms of the amount of computation required for the task. When the amount of computation required for the task is C and the total running time of the device is T, the total CPU energy consumption E cpu Calculated as E cpu =cf 2 C+P no_load T For screen display power consumption, use E screen Indicates that it is set as a constant during the operation of the mobile terminal; at the same time, E screen It also represents the static power consumption of all mobile terminals including the screen display power consumption; For smart terminals, their energy consumption includes CPU energy consumption E cpu , network transmission energy consumption E tran , screen display power consumption E screen ; Then the total energy consumption E of the intelligent terminal is expressed as E = E cpu +E tran +E screen For intelligent terminal i∈I, its total energy consumption is E i , the remaining battery power is B i , then the remaining battery life is t bai-i for If the user wants to extend the average expected battery life of the device, he needs to maximize the function under the conditions of satisfying the constraints C1, C2, and C3. T sum =∑ i∈I t bat-i ,i∈I T sum The sum of the battery life of all devices; If the user wants to give priority to extending the battery life of low-power devices, he needs to maximize the function under the conditions of satisfying the constraints C1, C2, and C3. T min =min({t bat-i |i∈I}) T min To calculate the battery life of the device that runs out of power the fastest during the uninstall process; Users can also choose to optimize individual battery life with a certain weight ratio, or choose to optimize battery life and overall power consumption with a certain weight ratio; users can choose to comprehensively optimize task execution time and device power consumption, all of which must be optimized under the conditions of satisfying restrictions C1, C2, and C3; users can also choose to consider whether to uninstall newly added devices separately after the HomeCloud smart control center has obtained the optimal solution, or choose to let the HomeCloud smart control center decide whether to uninstall them separately.

Images