Mobile crowd sensing task matching method, device and equipment and storage medium
Through the two-stage trading mechanism and intelligent algorithm, the multi-dimensional demand and limited resource problems in MCS task matching are solved, efficient matching between tasks and workers and fair reward distribution are achieved, and workers' enthusiasm is improved.
Patent Information
- Application Number
- CN202410809273.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-21
- Publication Date
- 2024-11-19
AI Technical Summary
Mobile group intelligence perception (MCS) task matching is in the real world with multidimensional needs and limited resources, resulting in poor matching between tasks and workers, and it is difficult to distribute rewards fairly, reducing workers' enthusiasm.
Using a two-stage trading mechanism, using the enhanced ant colony optimization algorithm (EACO-3PTR) to perform path pre-planning and task recommendation in the futures trading stage, the dynamic planning algorithm selects the optimal workers, and uses the deep Q network model (DQN based on priority experience replay -PER) Adjust the task sequence during the spot trading stage, reduce interaction costs and risks, and improve matching effect.
It effectively reduces the interaction cost between tasks and workers, steadily allocates tasks, and fairly distributes rewards, improves workers' enthusiasm, and improves task matching effect.
Smart Images

Figure CN118982157A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of mobile crowd intelligence sensing technology, and in particular to a mobile crowd intelligence sensing task matching method, device, equipment and storage medium. Background Art
[0002] Mobile crowd sensing (MCS) represents a popular innovative way of information collection, which uses ubiquitous IoT devices equipped with advanced sensors and processors to collect heterogeneous data. It provides a cost-effective and reliable sensing paradigm, providing important support in a variety of applications such as smart transportation, environmental monitoring, mobile health care, and urban public management.
[0003] In order to better attract various IoT devices (also known as workers) to participate in the execution of MCS tasks, it is crucial to design appropriate incentive mechanisms, especially in the case of dealing with workers' selfishness and limited resources. For example, a worker with his own local workload is usually unwilling to provide data services to other tasks for free. To this end, a service trading market can be established on the MCS network to promote data sharing among workers (workers can get paid while contributing data) and support MCS task owners (they can pay workers for the data resources they obtain for MCS tasks). However, the MCS network environment in the real world is not static, but full of uncertainties. At the same time, the multi-dimensional requirements of MCS tasks (such as data freshness, location of MCS tasks, perception time window of MCS tasks, cost of MCS task owners, etc.) also bring great challenges to recruiting workers for MCS tasks.
[0004] In view of this, how to improve the MCS task matching effect has become a technical problem that needs to be solved urgently. Summary of the invention
[0005] Embodiments of the present invention provide a mobile crowd-sensing task matching method, apparatus, device and storage medium.
[0006] In a first aspect, an embodiment of the present invention provides a mobile crowd sensing task matching method, the method comprising:
[0007] S1, initialization and parameter definition of the mobile crowd sensing MCS network;
[0008] S2, in the futures trading stage, for any worker, he determines the MCS task sequence with the maximum expected benefit based on the multi-dimensional attributes of each MCS task and his own personalized attributes, and sends requests to each MCS task in the MCS task sequence, and announces his own capabilities and required remuneration;
[0009] S3, for any MCS task, it converts the worker selection problem into a knapsack problem according to the ability, required compensation and expected benefits of the workers who send requests to it, and selects the worker who brings the maximum expected benefit, and sends the corresponding selection results to each worker who sends requests to it;
[0010] S4, for any worker, when faced with the selection result sent by the MCS task, if the selection result is accepted by the MCS task, the required reward remains unchanged; if the selection result is rejected, the worker reduces the required reward according to the expected cost of executing the MCS task, re-determines the MCS task sequence with the maximum expected benefit, and sends requests to each MCS task in the MCS task sequence;
[0011] S5, repeat S2 to S4 until the remuneration required by each worker no longer changes, at which point each MCS task signs a forward contract with the worker it selects.
[0012] In some implementations of the first aspect, S1 includes:
[0013] Divide each transaction into |T| time slots, T = {1, 2, ..., t, ..., |T|};
[0014] Setting MCS tasks i ∈MCS task set S, consisting of 6 tuples It indicates that, Respectively represent s i The start time slot and the end time slot of B together form a time window; i Represents MCS tasks i budget; Represents MCS tasks i Expected quality of service; MCS tasks i The location of Given, and Represents MCS tasks s i The longitude and latitude of D i Represents the tasks provided by each worker to MCS i The data size;
[0015] Set worker w j ∈ Worker set W, consisting of 6-tuples It indicates that, represents the cost of collecting data per time slot; Indicates the cost of transmitting data per time slot; represents the cost of traveling to the task location in each time slot; f j Indicated by worker wj The size of the data collected in each time slot; v j Represents worker w j Movement speed of workers w j The position at time slot t is given by express, and Represents workers w j The longitude and latitude at time slot t.
[0016] In some implementations of the first aspect, the algorithm used in S1 to determine the MCS task sequence with the maximum expected benefit is an enhanced ant colony optimization algorithm EACO-3PTR for path pre-planning and task recommendation.
[0017] In some implementations of the first aspect, the algorithm used in S3 to select the worker that brings the maximum expected benefit is a dynamic programming algorithm.
[0018] In some possible implementations of the first aspect, S5 further includes:
[0019] Each worker uses his or her corresponding MCS task sequence as the order and route in which he or she performs tasks.
[0020] In some implementations of the first aspect, the method further includes:
[0021] S6, in the spot trading stage, for any worker, it decides whether to continue to perform the current task or give up the current task according to the current environment and select the next task in the MCS task sequence;
[0022] S7, repeat S6 until the last time slot ends or the last task of the MCS task sequence is completed or abandoned.
[0023] In some implementations of the first aspect, in S6, whether to continue executing the current task or abandon the current task is determined according to the current environment, and the algorithm used to select the next task in the MCS task sequence is a deep Q network model DQN-PER based on prioritized experience replay.
[0024] In a second aspect, an embodiment of the present invention provides a mobile crowd intelligence sensing task matching device, the device comprising:
[0025] Initialization module, used for S1, to initialize the mobile crowd sensing MCS network and define parameters;
[0026] The determination module is used for S2. In the futures trading stage, for any worker, it determines the MCS task sequence with the maximum expected benefit based on the multi-dimensional attributes of each MCS task and its own personalized attributes, and sends requests to each MCS task in the MCS task sequence according to the MCS task sequence, and announces its own capabilities and required remuneration;
[0027] The selection module is used in S3. For any MCS task, it converts the worker selection problem into a knapsack problem according to the ability, required compensation and expected benefits of the workers who send the request to it, and selects the worker with the maximum expected benefit, and sends the corresponding selection result to each worker who sends the request to it;
[0028] The analysis module is used in S4. For any worker, when facing the selection result sent by the MCS task, if the selection result is accepted by the MCS task, the required reward remains unchanged. If the selection result is rejected, the required reward is reduced according to the expected cost of executing the MCS task, and the MCS task sequence with the maximum expected benefit is re-determined, and requests are sent to each MCS task in the MCS task sequence according to the MCS task sequence;
[0029] The signing module is used in S5 to repeat S2 to S4 until the remuneration required by each worker no longer changes. At this time, each MCS task signs a forward contract with the worker it selects.
[0030] In a third aspect, an embodiment of the present invention provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described above.
[0031] In a fourth aspect, an embodiment of the present invention provides a non-transitory computer-readable storage medium storing computer instructions, where the computer instructions are used to cause a computer to execute the method described above.
[0032] In the embodiments of the present invention, the interaction cost between MCS tasks and workers can be effectively reduced, and MCS tasks can be stably allocated to improve the MCS task matching effect. At the same time, rewards can be fairly allocated to improve the enthusiasm of workers.
[0033] It should be understood that the contents described in the summary of the invention are not intended to limit the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are provided for a better understanding of the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:
[0035] Figure 1 A flow chart of a mobile crowd-sensing task matching method provided by an embodiment of the present invention is shown;
[0036] Figure 2 A specific application schematic diagram of a mobile crowd intelligence sensing task matching method provided by an embodiment of the present invention is shown;
[0037] Figure 3 The figure shows the main interaction diagram between workers and tasks in the first trading stage, namely the futures trading stage, provided by an embodiment of the present invention;
[0038] Figure 4 A detailed architecture diagram of the DQN-PER algorithm in the second trading stage, i.e., the spot trading stage, provided by an embodiment of the present invention is shown;
[0039] Figure 5 A schematic diagram showing a comparison of the number of interactions between the mobile crowd-sensing task matching method provided by an embodiment of the present invention and the existing matching algorithm when achieving a match;
[0040] Figure 6 A schematic diagram showing a comparison of the running time of the mobile crowd intelligence sensing task matching method provided by an embodiment of the present invention and the existing matching algorithm when achieving matching;
[0041] Figure 7 A schematic diagram showing a comparison of the service quality obtained by the MCS task when the mobile crowd intelligence sensing task matching method provided by an embodiment of the present invention and the existing matching algorithm are matched;
[0042] Figure 8 A structural diagram of a mobile crowd-sensing task matching device provided by an embodiment of the present invention is shown;
[0043] Fig. 9 A block diagram of an exemplary electronic device capable of implementing an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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.
[0045] In addition, the term "and / or" in the present invention is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects before and after are in an "or" relationship.
[0046] In order to solve the technical problems in the background technology, the embodiments of the present invention provide a mobile crowd intelligence sensing task matching method, device, equipment and storage medium. In conjunction with the accompanying drawings, the mobile crowd intelligence sensing task matching method, device, equipment and storage medium provided by the embodiments of the present invention are described in detail through specific embodiments.
[0047] Figure 1 A flow chart of a mobile crowd-sensing task matching method provided by an embodiment of the present invention is shown. Figure 1 As shown, the mobile crowd intelligence sensing task matching method 100 may include the following steps:
[0048] S1, initialize the MCS network and define parameters.
[0049] S2. In the futures trading stage, for any worker, he determines the MCS task sequence with the maximum expected benefit based on the multidimensional attributes of each MCS task and his own personalized attributes, sends requests to each MCS task in the MCS task sequence, and announces his own capabilities and required remuneration.
[0050] S3, for any MCS task, it converts the worker selection problem into a knapsack problem based on the capabilities of the workers who send requests to it, the required remuneration, and the expected benefits, and selects the workers who bring the maximum expected benefits, and then sends the corresponding selection results to each worker who sends a request to it.
[0051] S4, for any worker, when faced with the selection result sent by the MCS task, if the selection result is accepted by the MCS task, the required remuneration remains unchanged; if the selection result is rejected, the worker reduces the required remuneration based on the expected cost of executing the MCS task, re-determines the MCS task sequence with the maximum expected benefit, and sends requests to each MCS task in the MCS task sequence according to the MCS task sequence.
[0052] S5, repeat S2 to S4 until the remuneration required by each worker no longer changes. At this time, each MCS task signs a forward contract with the worker it selects, and each worker uses its corresponding MCS task sequence as the order and route for performing its tasks.
[0053] S6, in the spot trading stage, for any worker, it decides whether to continue to perform the current task or abandon the current task according to the current environment and select the next task in the MCS task sequence.
[0054] S7, repeat S6 until the last time slot ends or the last task of the MCS task sequence is completed or abandoned.
[0055] In order to facilitate further understanding, the above content is described in detail below in conjunction with specific embodiments:
[0056] like Figure 2 As shown, this embodiment provides a mobile crowd-sensing task matching method based on two-stage transactions that takes into account the dynamics of the real-world MCS network, as well as the geographical location, time window, data freshness and other multi-dimensional requirements of the task. While controlling risks, it can effectively reduce the interaction cost between tasks and workers, and can efficiently and stably allocate tasks, fairly allocate rewards, and improve the enthusiasm of workers. The specific steps are as follows:
[0057] Step 1: Initialize the MCS network and define parameters.
[0058] Divide each transaction into |T| time slots, T = {1, 2, ..., t, ..., |T|}.
[0059] Setting MCS tasks i ∈MCS task set S, consisting of 6 tuples It indicates that, Respectively represent s i The start time slot and the end time slot of B together form a time window; i Represents MCS tasks i of budgets to limit the number of workers that can be hired; Represents MCS tasks i The expected quality of service (e.g., the required data freshness); MCS tasksi The location of Given, and Represents MCS tasks s i The longitude and latitude of D i Represents the tasks provided by each worker to MCS i The data size (e.g., bits).
[0060] Set worker w j ∈ Worker set W, consisting of 6-tuples It indicates that, represents the cost of collecting data per time slot; Indicates the cost of transmitting data per time slot; represents the cost of traveling to the task location in each time slot; f j Indicated by worker w j The size of the data collected in each time slot (e.g., bits); v j Represents worker w j The movement speed of workers w j The position at time slot t is given by <t>)express, and Represents workers w j The longitude and latitude at time slot t.
[0061] Step 2: Calculation of benefits and costs of tasks and workers.
[0062] MCS Missions i and workers j The Euclidean distance at time slot t is calculated as Therefore, at time slot t, worker w j Move to MCS Tasks i The time it takes is The corresponding moving cost can be calculated as:
[0063]
[0064] Worker j The time spent collecting data can be expressed as The corresponding perceived cost is Worker j The time consumed in transmitting the sensing data is The corresponding transmission cost is Where W is the wireless transmission bandwidth, MCS tasks i The signal-to-noise ratio of the received wireless data. Therefore, the cost can be expressed as:
[0065]
[0066] Among them, t ini It's a worker j Start heading to MCS missions i The time slot at the beginning of the worker w j Complete MCS missions i The time slot consumed can be expressed as:
[0067]
[0068] in, Represented as worker w j Start heading to MCS missions i The time slot consumed by the delay event during the journey is related to two uncertain factors: (1) the worker w j Whether there will be a delay event α in each time slot on the way to the task location follows the Bernoulli distribution, α~B{(0,1),{a,1―a}}. If α=1, it means that worker w j Go to MCS Missions i A delay event occurs at a certain time slot in the journey, and vice versa. (2) The time spent on each delay event follows a discrete uniform distribution but
[0069] Since each MCS task i There is a time window The worker w j May encounter delay events, workers w j MCS missions may not be completed i Therefore, we use β i,j To describe the actual transaction of workers w j Have you completed the MCS tasks? i , if β i,j = 1, then in actual transactions, worker w j Completed MCS missions i , if β i,j = 0, then in actual transactions, worker w j Abandoned MCS missions i . Therefore, the worker's benefit can be expressed as:
[0070]
[0071] in, It's a worker j The set of accepted tasks, p i,j It means w j To MCS tasks i The required reward, q i,j It represents the worker w j Abandon MCS tasks i The compensation to be paid when It represents the worker w j Option to abandon MCS tasks i Previously, MCS tasks have been i The cost incurred.
[0072] Considering the importance of freshness in MCS data, AoI is used here as a key evaluation of the quality of service obtained by a task. In particular, the life cycle of data usually starts when a worker collects / sensed the data and ends when the data is delivered to the task owner. In this process, AoI will increase over time. Let Represents worker w j Start tasks for MCS i The time slots when collecting data (each time slot generates a size of f j Here it is assumed that once a worker completes the overall data collection for a task, the data is transferred to the task. Then, the worker w j Provides tasks to MCS i The AoI of the sensing data can be defined as:
[0073]
[0074] Therefore, workers w j MCS tasks i The average AoI of the overall sensing data contributed by can be calculated as Therefore, suppose Describe MCS tasks i Accepting MCS missions i Workers Set The benefits obtained: Therefore, MCS tasks i The utility of is composed of two key components: (1) the quality of service obtained; and (2) the remaining budget (the larger the remaining budget, the lower the cost), as shown below:
[0075]
[0076] in, and is a positive weighting coefficient.
[0077] Step 3: Futures trading stage.
[0078] like Figure 3 As shown in the figure, a futures trading-driven stable matching and pre-path-planning mechanism (FT-SM3P) is developed here to achieve efficient and stable matching between workers and time-dependent tasks, thereby obtaining a set of forward contracts between them to guide future transactions and the paths that individual workers may follow. Specifically, the two key algorithms of FT-SM3P are: (1) enhanced ant colony optimization for path pre-planning and task recommendation (EACO-3PTR), which takes into account different task requirements, the payment that workers can obtain, and the cost of performing tasks, and aims to recommend tasks and corresponding pre-planned paths to workers; (2) futures trading-driven many-to-many matching (FT-M2M), which is used to enable long-term contracts between workers and tasks to achieve stable MCS services.
[0079] Inspired by the ant colony optimization algorithm, the EACO-3PTR algorithm is developed here, taking into account the different characteristics of tasks (e.g., different time windows and locations) and workers (e.g., various costs). The algorithm recommends appropriate task sequences (i.e., tasks and pre-planned paths) to each worker to maximize the worker's expected utility. j and all the tasks as vertices and edges between them, here we construct a graph consisting of G j (V j , E j ) is a directed complete graph represented by j = {v 0 ,v 1 ,v 2 ,...,v |S| } represents the vertex set (v 0 Represents worker w j , v 1 ,v 2 ,...,v |S| represents a task), and for ease of analysis, it is indexed by m and n. j ={(v m , v n )|v m ,v n ∈v j , m≠n} is an edge set. Please note that v 0 The location of Indicates that v 0 ,v 1 ,v 2 ,...,v |S| The location is V j The position of can therefore be expressed as
[0080] In the EACO-3PTR algorithm, the set To represent the collection of ants. Each ant Simulated Worker j The moving trajectory, that is, from v 0 It considers comprehensive information such as the current time slot, the time window of the task, the location of the vertex, the expected time slot required to reach each task, and the expected utility when reaching each task, and then selects a feasible vertex for the next visit (i.e., the current time slot plus the expected travel time from the current location to the target vertex, earlier than the end time of the task time window of the target vertex). This process will continue until there are no applicable vertices to visit. The state transition rule of the EACO-PTR algorithm, namely, the ant From the task v m Move to v n The probability of (m≠n≠0) can be defined as
[0081]
[0082] And if but Among them, ε 1 ,ε 2 ,ε 3 ,ε 4 is a positive weight coefficient. And η m,n Indicates v m to v n The distance between n Represents MCS tasks i At the vertex v n The time window width in ). In addition, wait n represents the waiting time of the ant arriving early (e.g., before the task starts), where also, It's an ant In vertical v m The set of remaining feasible tasks that can be completed on . In addition, τ m,n Expressed as from v m to v n When the ant colony finds a path, the pheromones along the edge will be updated to:
[0083] τ m,n ←(1―θ)τ m,n +Δτ m,n (8)
[0084] Among them, (0<θ<1) is the pheromone evaporation coefficient, Δτ m,n is the pheromone deposited by the ants and can be calculated as:
[0085]
[0086] in, Represents the vertex v m Move to v n The worker w in the task j Expected utility. max is the path with the maximum utility obtained from all ants (e.g., w j The vector of completed tasks), that is
[0087] As an example, the key steps of the EACO-3PTR algorithm are as follows:
[0088] (1) Initialization: Based on information about workers and tasks (e.g., their locations, task time windows, worker rewards, and the expected time and energy required to perform tasks), the EACO-3PTR algorithm initializes a task for each worker w. j Construct a directed complete graph G j (V j , E j ).
[0089] (2) Path exploration: In each iteration, the ants at each vertex Multiple factors are evaluated, including: the current time, the expected travel time to the next unvisited vertex, the expected time to perform the task, and the time window. This evaluation is done to determine whether the task can be completed within the specified time frame, thereby obtaining and Then, according to From the collection The next vertex (task) is selected in . This iterative process continues until there are no more executable vertices.
[0090] (3) Pheromone update: Then, compare different vertex sequences Expected utility of association Among them, the path with the largest expected utility (i.e., M max ) to update pheromones.
[0091] (4) Optimal path output: After all iterations are completed, the path M with the highest expected utility can be determined. max As a worker j Accordingly, the set of tasks recommended along this path is the recommended worker w j Selected Tasks That is, the task sequence.
[0092] Further, following the pre-planned path, here is the FT-M2M algorithm, which promotes stable and efficient matching between tasks and workers, and encourages them to sign forward contracts. The details are as follows:
[0093] (1) Initialization: Each worker w j The requested payment is set to Y(w j ) contains worker w j Task of interest, Y(s i ) involves MCS tasks i Temporary selection of workers.
[0094] (2) Workers send requests: In each round k, each worker w j First from (According to its expected utility for worker w j The optimal task vector selected is derived by the EACO-3PTR algorithm) and records it in Y(w j ). Then, worker w j To Y(w j ) sends proposals to each task in the i,j <k>, complete MCS tasks i The probability (i.e., β i,j ), and the expected quality of service (E[Q i,j ]).
[0095] (3) Worker selection in terms of tasks: After collecting information from workers, each MCS task s i Solve a 0-1 knapsack problem, which can usually be solved by dynamic programming, and determine a set of temporary workers (for example, the set Y (s i )) can be used in budget B i Then, each MCS task s i Reports its decision on worker selection to all workers in the current round.
[0096] (4) Worker decision-making: After obtaining each MCS task s i ∈Y(w j ) decision, workers w j Check the following conditions:
[0097] Condition 1: If worker w j MCS tasks i Accept or currently require payment i,j <k>Equal to its cost c i,j , whose payments remain unchanged;
[0098] Condition 2: If worker w j MCS tasks i Refusal and the payment required i,j <k>Still can cover its cost c i,j , it will be s in the next round i Reduce the required payment, while avoiding disutility, as follows:
[0099] p i,j<k +1>=max{p i,j <k>―Δp j ,c i,j } (10)
[0100] (4) Repeated iteration: If from (k―1) th It's K's turn th If all required payoffs remain unchanged, the matching will terminate in round k. Otherwise, the algorithm repeats (2) to (4) in the next round.
[0101] (6) Risk analysis: When the algorithm terminates, each task and worker will conduct a risk assessment on the matching results and choose whether to sign a long-term contract. Three risk controls are considered here:
[0102] (1) The risk of workers obtaining unsatisfactory utility: Each worker serves the MCS task Workers j ∈W faces the risk of obtaining unsatisfactory utility in actual transactions (for example, when U W (w j ,s i ) becomes negative). This risk is defined as the utility U W (w j ,s i ) is lower than the tolerable value u min The probability is expressed as:
[0103]
[0104] Among them, u min is a positive value close to 0.
[0105] (2) Risk of not being able to complete the task: Due to possible delay events, workers w j Complete MCS mission This risk is defined as
[0106]
[0107] (3) Similar to workers, task owners also face risks due to market dynamics: MCS tasks i This risk is mainly related to the failure to achieve the expected quality of service, as shown below:
[0108]
[0109] When the above three risks are controlled within a reasonable range, workers and tasks choose to sign long-term contracts.
[0110] Step 4: Spot trading stage.
[0111] Faced with MCS dynamic networks and uncertainties (e.g., delayed events), workers need to choose whether to abandon the current task or continue to execute the current task to maximize their utility through a deep Q-network with prioritized experience replay-enhanced dynamic route optimization (DQNP-DRO). Next, the DQNP-DRO algorithm will be introduced.
[0112] The problem studied here is represented as a Markov decision process. The deep Q-network (DQN) provides a useful reinforcement learning technique that combines Q-learning with deep neural networks, benefiting from their online learning and function approximation capabilities. Accordingly, the DQNP-DRO algorithm proposed here implements prioritized experience replay (PER) on the DQN model.
[0113] like Figure 4 As shown, the dynamic MCS network considered here is conceptualized as an environment with multiple agents. At the same time, each worker is regarded as a DQN agent, which decides whether to continue the current task or to start from the task vector according to the observed environment state. Select the next task. The agent's goal is to Next Action To find the optimal strategy π, so as to maximize the long-term cumulative reward. The optimal state-action value can be found by the Q function based on the state-action pair, defined as:
[0114]
[0115] Where χ is the discount factor (0≤χ≤1), is an agent that takes action In Status The reward that can be obtained. Q learning uses a Q table to store Q values, and the online update rule of the Q value and the learning rate lr (0≤lr≤1) are given as follows:
[0116]
[0117] The state, action, and reward functions of the DQNP-DRO algorithm proposed here are defined as follows:
[0118] State: The state consists of five elements: (1) the worker’s current location; (2) the current time slot; (3) the target task; (4) the information required to perform the task (e.g., the sequence in which tasks are to be performed, the location and time window of the tasks, and the penalty for abandoning the task); and (5) the distance from the worker to the current target task.
[0119] Action: Actions consider two main factors: (1) continue to execute the currently selected task, (2) abandon the current task and select the next task to follow the predetermined path.
[0120] Reward function: The reward function considers four conditions: (1) If a worker chooses to continue the current task while on the way to a task, a negative reward of movement loss is given (2) If a worker plans to change his current task, he may incur a negative reward due to default (q i,j ); (3) After arriving at the mission location, the data collection and transmission costs will lead to negative rewards (4) Completing the task will earn you a positive reward (p i,j ).
[0121] Different from Q learning, DQN uses a deep neural network to approximate the optimal state-action value Q function, namely θ represents a set of parameters of the neural network. The training process of DQNP-DRO includes the following steps:
[0122] (1) Initialization: First, initialize the relevant parameters, where time slot t = 0 and for each worker w j ∈W initializes the DQN network.
[0123] (2) DQN model training: Before taking action, worker w j First, it observes the current state of the environment. Based on this state, it takes action through the Q value generated by DQN, obtains the corresponding reward, and updates the state to Next, the information related to time slot t+1 is stored in a memory buffer and the DQN is updated using a prioritized experience replay strategy. This process is repeated until all tasks of the forward contract are completed or abandoned.
[0124] (3) Repeat iterations: (1) to (2) are repeated until the reward curve converges.
[0125] The following can be combined Figure 5-7 The mobile crowd-sensing task matching method 100 provided by the embodiment of the present invention, i.e., the two-stage transaction stable matching algorithm, is shown to have different effects in terms of the number of interactions, running time, and service quality compared with the greedy matching algorithm, the traditional single-stage many-to-many matching algorithm, and the random matching algorithm. It can be seen that according to the embodiment of the present invention, at least the following technical effects are achieved:
[0126] Here, the task matching of mobile crowd intelligence perception is divided into two stages: futures trading stage and spot trading stage. First, in the futures trading stage, by comprehensively considering the multi-dimensional requirements of tasks (perception data freshness, location, perception time window) and dynamic environment, combined with the specific capabilities and location information of workers, based on historical statistical information and risk control, tasks and workers are stably matched, and the paths of workers are pre-planned. According to the matching results, workers and tasks sign forward contracts; secondly, in the spot trading stage, facing the dynamic network environment, the DQN-PER algorithm is proposed to assist workers to complete tasks according to forward contracts and maximize workers' benefits and task completion rates. This method can take into account the dynamics of the real-world MCS network, as well as the multi-dimensional requirements of tasks such as geographical location, time window, and data freshness. While controlling risks, it can effectively reduce the interaction cost between tasks and workers, and can efficiently and stably allocate tasks, fairly allocate rewards, and improve workers' enthusiasm.
[0127] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described order of actions, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present invention.
[0128] The above is an introduction to a method embodiment. The following is a further explanation of the solution of the present invention through an apparatus embodiment.
[0129] Figure 8 FIG. 4 shows a structural diagram of a mobile crowd-sensing task matching device provided by an embodiment of the present invention. Figure 8 As shown, the mobile crowd intelligence sensing task matching device 800 may include:
[0130] The initialization module 810 is used for S1 to initialize the mobile crowd sensing MCS network and define parameters.
[0131] Determination module 820, used for S2, in the futures trading stage, for any worker, it determines the MCS task sequence with the maximum expected benefit based on the multi-dimensional attributes of each MCS task and its own personalized attributes, and sends requests to each MCS task in the MCS task sequence according to the MCS task sequence, and announces its own capabilities and required remuneration.
[0132] The selection module 830 is used for S3. For any MCS task, it converts the worker selection problem into a knapsack problem based on the capabilities of the workers who send requests to it, the required remuneration, and the expected benefits, and selects the workers who bring the maximum expected benefits, thereby sending the corresponding selection results to each worker who sends a request to it.
[0133] Analysis module 840 is used for S4. For any worker, when facing the selection result sent by the MCS task, if the selection result is accepted by the MCS task, the required remuneration remains unchanged. If the selection result is rejected, the required remuneration is reduced according to the expected cost of executing the MCS task, and the MCS task sequence with the maximum expected benefit is re-determined, and requests are sent to each MCS task in the MCS task sequence according to the MCS task sequence.
[0134] The signing module 850 is used for S5, and S2 to S4 are repeated until the remuneration required by each worker no longer changes. At this time, each MCS task signs a forward contract with the worker it selects.
[0135] In some embodiments, the signing module 850 is further configured to:
[0136] Each worker uses his or her corresponding MCS task sequence as the order and route in which he or she performs tasks.
[0137] In some embodiments, the mobile crowd intelligence sensing task matching device 800 further includes:
[0138] The decision module 860 is used in S6. In the spot trading stage, for any worker, it decides whether to continue to perform the current task or give up the current task according to the current environment and selects the next task in the MCS task sequence.
[0139] Processing module 870, used for S7, repeats S6 until the last time slot ends or the last task of the MCS task sequence is completed or abandoned.
[0140] Understandably, Figure 8 Each module / unit in the mobile crowd intelligence sensing task matching device 800 shown in the figure has the function of realizing Figure 1 The functions of each step in the mobile crowd sensing task matching method 100 shown in the figure can achieve its corresponding technical effects, and for the sake of brevity, they will not be repeated here.
[0141] Fig. 9 A block diagram of an exemplary electronic device capable of implementing an embodiment of the present invention is shown. Electronic device 900 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 900 may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown in the present invention, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required in the present invention.
[0142] like Fig. 9 As shown, the electronic device 900 may include a computing unit 901, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the electronic device 900 may also be stored. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.
[0143] Multiple components in the electronic device 900 are connected to the I / O interface 905, including: an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the electronic device 900 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.
[0144] The computing unit 901 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 901 performs the various methods and processes described above, such as method 100. For example, in some embodiments, the method 100 may be implemented as a computer program product, including a computer program, which is tangibly contained in a computer-readable medium, such as a storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the computing unit 901, one or more steps of the method 100 described above may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured to perform the method 100 in any other appropriate manner (e.g., by means of firmware).
[0145] The various embodiments described above in the present invention can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs, which may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general programmable processor, which may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0146] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.
[0147] In the context of the present invention, a computer-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a computer-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0148] It should be noted that the present invention also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute method 100 and achieve the corresponding technical effect achieved by the embodiments of the present invention executing the method. For the sake of concise description, they are not repeated here.
[0149] In addition, the present invention also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method 100 is implemented.
[0150] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in the present invention can be achieved, and the present invention is not limited here.
[0151] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.< / k> < / k> < / k> < / k> < / t>
Claims
1. A mobile crowd sensing task matching method, It is characterized in that The method comprises: S1, initialization and parameter definition of the mobile crowd sensing MCS network; S2, in the futures trading stage, for any worker, he determines the MCS task sequence with the maximum expected benefit based on the multi-dimensional attributes of each MCS task and his own personalized attributes, and sends requests to each MCS task in the MCS task sequence, and announces his own capabilities and required remuneration; S3, for any MCS task, it converts the worker selection problem into a knapsack problem according to the ability, required compensation and expected benefits of the workers who send requests to it, and selects the worker who brings the maximum expected benefit, and sends the corresponding selection results to each worker who sends requests to it; S4, for any worker, when faced with the selection result sent by the MCS task, if the selection result is accepted by the MCS task, the required reward remains unchanged; if the selection result is rejected, the worker reduces the required reward according to the expected cost of executing the MCS task, re-determines the MCS task sequence with the maximum expected benefit, and sends requests to each MCS task in the MCS task sequence; S5, repeat S2 to S4 until the remuneration required by each worker no longer changes, at which point each MCS task signs a forward contract with the worker it selects.
2. The method according to claim 1, It is characterized in that The S1 includes: Divide each transaction into |T| time slots, T = {1, 2, ..., t, ..., |T|}; Setting MCS tasks i ∈MCS task set S, consisting of 6 tuples It indicates that, Respectively represent s i The start time slot and the end time slot of B together form a time window; i Represents MCS tasks i budget; Represents MCS tasks i Expected quality of service; MCS tasks i The location of Given, and Represents MCS tasks s i The longitude and latitude of D i Represents the tasks provided by each worker to MCS i The data size; Set worker w j ∈ Worker set W, consisting of 6-tuples It indicates that, represents the cost of collecting data per time slot; Indicates the cost of transmitting data per time slot; represents the cost of traveling to the task location in each time slot; f j Indicated by worker w j The size of the data collected in each time slot; v j Represents worker w j Movement speed of workers w j The position at time slot t is given by express, and Represents workers w j The longitude and latitude at time slot t.
3. The method according to claim 1, It is characterized in that The algorithm used in S1 to determine the MCS task sequence with the maximum expected benefit is the enhanced ant colony optimization algorithm EACO-3PTR for path pre-planning and task recommendation.
4. The method according to claim 1, It is characterized in that The algorithm used in S3 to select the worker that brings the maximum expected benefit is a dynamic programming algorithm.
5. The method according to claim 1, It is characterized in that The S5 further includes: Each worker uses his or her corresponding MCS task sequence as the order and route in which he or she performs tasks.
6. The method according to claim 5, It is characterized in that The method further comprises: S6, in the spot trading stage, for any worker, it decides whether to continue to perform the current task or give up the current task according to the current environment and select the next task in the MCS task sequence; S7, repeat S6 until the last time slot ends or the last task of the MCS task sequence is completed or abandoned.
7. The method according to claim 6, It is characterized in that In S6, whether to continue executing the current task or abandon the current task is determined according to the current environment, and the algorithm used to select the next task in the MCS task sequence is the deep Q network model DQN-PER based on prioritized experience replay.
8. A mobile crowd-sensing task matching device, It is characterized in that The device comprises: Initialization module, used for S1, to initialize the mobile crowd sensing MCS network and define parameters; The determination module is used for S2. In the futures trading stage, for any worker, it determines the MCS task sequence with the maximum expected benefit based on the multi-dimensional attributes of each MCS task and its own personalized attributes, and sends requests to each MCS task in the MCS task sequence according to the MCS task sequence, and announces its own capabilities and required remuneration; The selection module is used in S3. For any MCS task, it converts the worker selection problem into a knapsack problem according to the ability, required compensation and expected benefits of the workers who send the request to it, and selects the worker with the maximum expected benefit, and sends the corresponding selection result to each worker who sends the request to it; The analysis module is used in S4. For any worker, when facing the selection result sent by the MCS task, if the selection result is accepted by the MCS task, the required reward remains unchanged. If the selection result is rejected, the required reward is reduced according to the expected cost of executing the MCS task, and the MCS task sequence with the maximum expected benefit is re-determined, and requests are sent to each MCS task in the MCS task sequence according to the MCS task sequence; The signing module is used in S5 to repeat S2 to S4 until the remuneration required by each worker no longer changes. At this time, each MCS task signs a forward contract with the worker it selects.
9. An electronic device, It is characterized in that The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.
10. A non-transitory computer-readable storage medium storing computer instructions, It is characterized in that The computer instructions are used to make a computer execute the method according to any one of claims 1 to 7.
Citation Information
Cited By
Remote sensing satellite area coverage method based on greedy search
CN121861506A
