UAM-assisted MEC cache system based on federal DDQN

Abstract

The invention discloses a UAM-assisted MEC caching system based on a federal DDQN, and the system comprises a UAM aircraft, and also comprises an unmanned plane, an edge server, and a macro base station. The unmanned aerial vehicle is used for carrying the edge server; the edge server is used for assisting the UAM aircraft in task processing in the air; and the macro base station is deployed on the ground and is used for assisting the UAM aircraft in task processing on the ground. According to the invention, for an air-ground edge calculation (MEC) technology in an urban air traffic (UAM) scene, the problems of limited terminal calculation capability and limited communication resources are mainly considered. The unmanned aerial vehicle can provide cache service and deployment of service content so as to meet the limitation requirements of the UAM aircraft on time delay and energy consumption. The algorithm shows excellent performance in the UAM cache assisted urban air traffic system. By dynamically adjusting resource allocation, the algorithm significantly reduces the overall time delay and energy consumption of the system, and effectively improves the operation efficiency of the system.

Description

Technical Field

[0001] The present invention relates to the field of wireless communications, and in particular to a UAM-assisted MEC caching system based on a federated DDQN. Background Art

[0002] With the explosive growth of mobile data traffic and the accelerated urbanization process, the demand for developing the low-altitude economy is increasing. Urban air mobility (UAM) is seen as a way to alleviate large-scale data processing and promote the development of the low-altitude economy. However, the current urban air mobility aircraft are limited by limited computing and storage capabilities, which makes it difficult to meet the needs of complex task processing. This bottleneck limits the stability and response speed of the UAM system in complex environments. In particular, when facing urgent tasks or intensive traffic needs, insufficient computing resources may directly affect the safety and reliability of the system. To overcome this obstacle, mobile edge computing (MEC) has emerged as a cutting-edge and viable solution. Despite the potential benefits of MEC in edge data computing, there are still certain challenges when edge servers are statically deployed on base stations. For example, in urban areas, the communication link between the base station and the UAM aircraft may be interfered by dense buildings, resulting in unstable communication and task processing. In addition, at large gatherings with a large number of users, a large number of tasks arriving at the same time make it difficult for the edge server to process tasks in a timely manner. Therefore, a more flexible edge server deployment strategy is urgently needed. In recent years, unmanned aerial vehicles (UAVs) have been widely used in MEC due to their flexible mobility, low price, and line-of-sight communication links. In the urban air traffic (UAM) scenario, drones equipped with edge servers can provide computing and storage services. With the assistance of drones, computing tasks from UAM aircraft can be flexibly offloaded to base stations or drones according to the current network status. In addition, in dynamic scenarios, protecting user privacy and obtaining the optimal global data caching strategy efficiently and at low cost are also key issues in UAM scenarios. Based on the above analysis, an algorithm combining federated learning and deep reinforcement learning is proposed. By constructing a three-dimensional deployment and resource allocation model for urban air traffic, a UAM-assisted MEC caching method and device based on federated DDQN is designed. Summary of the invention

[0003] The system proposed in this application uses a mobile edge computing server carried by a drone as a mobile edge node in an emergency scenario to effectively optimize computing resources, communication resources and offloading decisions to meet the latency and energy consumption requirements of UAM aircraft. This algorithm focuses on optimizing data processing efficiency in a dynamic and fast-response environment. By using drones equipped with edge servers to achieve reasonable offloading of UAM aircraft tasks and full utilization of computing resources, the algorithm aims to improve the data processing efficiency of urban air traffic scenarios. In addition, the design of the algorithm can also achieve more efficient task offloading and computing in emergency or difficult-to-access areas, while maintaining a high degree of operational flexibility and computing efficiency to meet the needs of the future intelligent society for efficient and real-time communication networks.

[0004] To achieve the above object, the present invention provides a UAM-assisted MEC caching system based on a federated DDQN, comprising: a UAM aircraft, characterized in that it also includes: a drone, an edge server and a macro base station;

[0005] The drone is used to carry the edge server;

[0006] The edge server is used to assist the UAM aircraft in performing mission processing in the air;

[0007] The macro base station is deployed on the ground to assist the UAM aircraft in performing mission processing on the ground.

[0008] Preferably, when the UAV and the UAM aircraft k perform data transmission, the downlink rate r k，n It is expressed as:

[0009]

[0010] Among them, σ 2 represents the variance of the additive white Gaussian noise in the wireless channel; W k,n represents the downlink bandwidth allocated by the UAV to UAM aircraft k, which depends on the spectrum resource allocation strategy of the system; N represents the number of UAM aircraft; h k,n represents the channel power gain between the UAV and the UAM aircraft; h k，j represents the channel gain between the jth UAM aircraft and the kth UAM aircraft; p represents the UAM transmission power;

[0011] UAV backhaul link rate r b，n It is expressed as:

[0012]

[0013] Among them, W b,n P represents the backhaul link bandwidth allocated to the macro base station connected to the drone;bs Indicates the base station transmission power; h b,n Represents the channel gain between the macro base station and the drone.

[0014] Preferably, when the macro base station communicates with the UAM aircraft k:

[0015]

[0016] Among them, W b,k represents the downlink frequency band resource allocated by the macro base station to UAM aircraft k, h b,k represents the channel gain between the macro base station and UAM aircraft k.

[0017] Optimized wireless transmission delay of drone content cache It is expressed as:

[0018]

[0019] Among them, hit i，k =1 indicates that the cache content hits, i.e. the service content has been cached in the drone; k，m represents the preference of aircraft k for content m; D k Represents the size of the cache content of the kth UAM aircraft.

[0020] Preferably, the UAM aircraft processes tasks in the following ways: performing calculations locally, offloading to the UAV for calculations, and offloading to a macro base station.

[0021] Preferably, when performing calculations locally, the computing power of the UAM aircraft is denoted as f k , assuming that the local unloading decision of UAM vehicle k is z k =1, otherwise z k =0; when UAM aircraft k offloads the task to the UAV When offloading tasks to macro base stations

[0022] If UAM vehicle k decides to perform the mission locally, then the delay T for processing the mission is k for:

[0023]

[0024] Among them, C k Indicates the CPU cycles required for the task.

[0025] The energy consumption of its task execution is E k It is expressed as:

[0026] E k =ρ k (fk ) 2

[0027] Among them, ρ k It is the energy consumption coefficient related to UAM aircraft, which depends on the CPU performance structure.

[0028] Preferably, when the UAM aircraft applies for content services to the drone, there are two situations. The first is when the drone has cached the required content services locally, the drone directly calculates the UAM aircraft task; the second is when the drone does not have cached the required services, the drone applies to the macro base station, and the macro base station processes the task.

[0029] Preferably, when the local directly applies for service from the macro base station, the transmission delay when the UAM aircraft k transmits the task to the macro base station is for:

[0030]

[0031] The device assigns the task to the macro base station to calculate the delay T b for:

[0032]

[0033] Among them, r b，k represents the transmission rate between UAM aircraft k and the macro base station; F b represents the average computing power of the macro base station; C k Indicates the CPU cycles required for the task; L k Indicates the size of the task.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] This invention focuses on the air-to-ground edge computing (MEC) technology in the urban air traffic (UAM) scenario, and focuses on the problems of limited terminal computing power and communication resources. UAVs can provide cache services and deployment of service content to meet the UAM aircraft's restrictions on latency and energy consumption. The algorithm shows excellent performance in the UAM cache-assisted urban air traffic system. By dynamically adjusting resource allocation, the algorithm significantly reduces the overall latency and energy consumption of the system, and effectively improves the system's operating efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0037] Figure 1 It is a schematic diagram of the system structure of the present invention;

[0038] Figure 2 Detailed schematic diagram of the system of the present invention. DETAILED DESCRIPTION

[0039] 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.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] like Figure 1 As shown, it is a schematic diagram of the system structure of this embodiment, including: UAM aircraft, also including: drone, edge server and macro base station; the drone is used to carry the edge server; the edge server is used to assist the UAM aircraft to perform task processing in the air; the macro base station is deployed on the ground, and is used to assist the UAM aircraft to perform task processing on the ground.

[0042] Specifically, in the Urban Air Mobility (UAM) system, unmanned aerial vehicles (UAVs) are equipped with edge servers to implement caching and computing of mobile multimedia services. UAVs fly in complex urban environments and dynamically adjust their flight paths to optimize the allocation of computing resources and the efficiency of task offloading. Macro base stations are deployed on the ground to support the connection of multiple UAM aircraft.

[0043] Specifically, macro base stations and other infrastructure are deployed on the ground side, and N UAM aircraft are deployed with three-dimensional coordinates (x k ,y k ,H UAM The UAV flies above the UAM aircraft at a fixed altitude H. The three-dimensional coordinates of the UAV are marked as (x u ,y u ,H+H UAM ).

[0044] The communication frequency band between the UAV and the UAM aircraft is W n , the communication frequency band between the base station and the UAM aircraft is W b , the backhaul link bandwidth between the drone and the base station is W e , in order to avoid interference, where Wn , W b With W e Mutually orthogonal.

[0045] The distance d between the UAV and the UAM aircraft k k for:

[0046]

[0047] Where H represents the vertical height of the UAV from the UAM aircraft.

[0048] The communication between the UAV and the UAM aircraft is transmitted using a wireless link, and the channel power gain is based on the free space path loss model and the random fading factor f k,n , expressed as follows:

[0049]

[0050] Where p represents the UAV transmission power; h k,n It represents the channel power gain between the UAV and the UAM aircraft.

[0051] When the UAV and UAM aircraft k transmit data, the downlink rate r k，n It can be expressed as:

[0052]

[0053] Among them, σ 2 represents the variance of the additive white Gaussian noise in the wireless channel; W k,n represents the downlink bandwidth allocated by the UAV to UAM aircraft k, which depends on the spectrum resource allocation strategy of the system; N represents the number of UAM aircraft; h k，j Denotes the channel gain between the j-th UAM aircraft and the k-th UAM aircraft.

[0054] UAV backhaul link rate r b，n It can be expressed as:

[0055]

[0056] Among them, W b,n P represents the backhaul link bandwidth allocated to the macro base station connected to the drone; bs Indicates the base station transmission power; h b,n Represents the channel gain between the macro base station and the drone.

[0057] When the macro base station communicates with the UAM aircraft k:

[0058]

[0059] Among them, W b,k represents the downlink frequency band resource allocated by the macro base station to UAM aircraft k, h b,k represents the channel gain between the macro base station and the UAM aircraft k; r b，k represents the transmission rate between UAM aircraft k and the macro base station.

[0060] The content cached by the drone is represented as a set C M , the drone can cache at most C complete content services. Definition is a binary variable representing the request of aircraft k for content m service. is a binary variable, It means that aircraft k applies for content m service from drone. is a binary variable, It means that aircraft k requests m content service from base station. The service content cached by drone and base station follows Zipf Law (Zipf Law, Zipf), (p m ) M×1 is the global popularity, which represents the probability distribution of all UEs requesting content m in the network. Let p nm is the popularity of content m under drone n. Considering p m =∑ n∈N p nm , where (p m ) M×1 Satisfies Zipf distribution:

[0061]

[0062] Among them, I m It is arranged in descending order according to the popularity of content m, and τ and β represent the platform factor and skewness factor respectively.

[0063] Considering p k,m =r k,m / R k ,m∈M, represents the preference of aircraft k for content m,∑ m∈M P k,m =1, where r k,m is the number of requests from aircraft k for content m, R k is the total number of requests from aircraft k in the network. In addition, this embodiment defines p k,m =p k,m a k is the preference of aircraft k for content m under the UAV, a k is the probability of association between aircraft k and drone, so the wireless transmission delay of content cache It is expressed as:

[0064]

[0065] Among them, hit i，k =1 indicates that the cache content hits, that is, the service content has been cached in the drone, and the task can be offloaded to the drone for calculation; D k represents the kth UAM aircraft.

[0066] UAM aircraft can choose three ways to process tasks: 1) perform calculations locally, 2) offload calculations to the UAV, and 3) offload to a macro base station.

[0067] The processed tasks are designed as a four-tuple {L, F, t, m}, where L represents the task size, F represents the CPU cycles required for calculation, t represents the maximum tolerable delay, and m represents the service content. This embodiment assumes that the UAM aircraft can choose to offload tasks to drones or edge macro base stations, or perform local calculations. The UAM aircraft can only offload its computing tasks to drones when the corresponding services are cached in the drone server.

[0068] (1) Local computing model

[0069] UAM aircraft have a certain computing power, which is expressed by CPU frequency, denoted as f k . Assume that the local unloading decision of UAM vehicle k is z k =1, otherwise z k = 0, also set UAM aircraft k to offload the task to the UAV When offloading tasks to macro base stations Consider a binary offload decision, so:

[0070]

[0071] If UAM vehicle k decides to perform the mission locally, then the delay T for processing the mission is k for:

[0072]

[0073] Among them, C k Indicates the CPU cycles required for the task.

[0074] The energy consumption of its task execution is E k It is expressed as:

[0075] E k =ρ k (f k ) 2

[0076] Among them, ρ kIt is the energy consumption coefficient related to UAM aircraft, which depends on the CPU performance structure.

[0077] (2) UAV calculation model

[0078] When a UAM aircraft applies for content services from a drone, there are two situations. The first is when the drone has cached the required content services locally, and the drone can directly calculate the UAM aircraft task. The second is when the drone does not have the required services cached, the drone applies to the macro base station, and the macro base station processes the task. The content needs to be transmitted to the macro base station. When the UAM aircraft assigns tasks to drones, it needs to consider the impact of the UAV's dynamic position on the calculation delay and communication delay. At the same time, due to the limited energy of drones, task allocation needs to comprehensively consider energy consumption and range limitations.

[0079] The transmission delay of UAM aircraft k transmitting the task to UAV for:

[0080]

[0081] The computational delay of UAM vehicle k assigning tasks to UAVs It is expressed as:

[0082]

[0083] Among them, F n , F b They represent the average computing power of UAV and macro base station respectively; L k Indicates the size of the task in bits; let hit k =1 means the drone has cache, hit k =0 means the drone has no cache.

[0084] (3) Macro Base Station Calculation Model

[0085] When the local area directly requests service from the macro base station, the transmission delay when the UAM aircraft k transmits the task to the macro base station for:

[0086]

[0087] Device i assigns the task to the macro base station to calculate the delay T b for:

[0088]

[0089] Since the computing power of the macro base station and the UAV is much larger than the local computing power of the UAM aircraft, the computing power of the UAV or the macro base station is ignored, and only the energy consumed by its task transmission is calculated. They can be expressed as:

[0090]

[0091] Therefore, the total delay of the UAV to perform the task can be: If the task is transmitted to the macro base station, the total delay of executing the task is: The total delay of task execution can be expressed as:

[0092]

[0093] The total energy consumption of task execution can be expressed as:

[0094]

[0095] Among them, p k represents the UAM aircraft transmission power vector, f k represents the allocation of computing resources, p m represents the popularity of service content, z k Indicates the local calculation of the UAM aircraft, Indicates application for drone unloading. Indicates application for macro base station task offloading. The detailed structure of this embodiment is as follows Figure 2 shown.

[0096] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A UAM-assisted MEC cache system based on federated DDQN, characterized in that: include: The UAM aircraft is characterized by further comprising: a drone, an edge server and a macro base station; The drone is used to carry the edge server; The edge server is used to assist the UAM aircraft in performing mission processing in the air; The macro base station is deployed on the ground to assist the UAM aircraft in performing mission processing on the ground.

2. The UAM-assisted MEC cache system based on federated DDQN according to claim 1 is characterized in that: When the UAV and UAM aircraft k transmit data, the downlink rate r k，n It is expressed as: Among them, σ 2 represents the variance of the additive white Gaussian noise in the wireless channel; W k,n represents the downlink bandwidth allocated by the UAV to UAM aircraft k, which depends on the spectrum resource allocation strategy of the system; N represents the number of UAM aircraft; h k,n represents the channel power gain between the UAV and the UAM aircraft; h k，j represents the channel gain between the jth UAM aircraft and the kth UAM aircraft; p represents the UAM transmission power; UAV backhaul link rate r b，n It is expressed as: Among them, W b,n P represents the backhaul link bandwidth allocated to the macro base station connected to the drone; bs Indicates the base station transmission power; h b,n Represents the channel gain between the macro base station and the drone.

3. The UAM-assisted MEC cache system based on federated DDQN according to claim 2 is characterized in that: When the macro base station communicates with the UAM aircraft k: Among them, W b,k represents the downlink frequency band resource allocated by the macro base station to UAM aircraft k, h b,k represents the channel gain between the macro base station and the UAM aircraft k; r b，k represents the transmission rate between UAM aircraft k and the macro base station.

4. The UAM-assisted MEC cache system based on federated DDQN according to claim 3 is characterized in that: Wireless transmission delay of drone content caching It is expressed as: Among them, hit i，k =1 indicates that the cache content hits, i.e. the service content has been cached in the drone; k，m represents the preference of aircraft k for content m; D k Represents the size of the cache content of the kth UAM aircraft.

5. The UAM-assisted MEC cache system based on federated DDQN according to claim 4 is characterized in that: The UAM aircraft processes tasks in the following ways: performing calculations locally, offloading calculations to drones, and offloading to macro base stations.

6. The UAM-assisted MEC cache system based on federated DDQN according to claim 5 is characterized in that: When computing locally, the computing power of the UAM aircraft is denoted as f k , assuming that the local unloading decision of UAM vehicle k is z k =1, otherwise z k =0; when UAM aircraft k offloads the task to the UAV When offloading tasks to macro base stations If UAM vehicle k decides to perform the mission locally, then the delay T for processing the mission is k for: Among them, C k Indicates the CPU cycles required for the task; The energy consumption of its task execution is E k It is expressed as: E k =ρ k (f k ) 2 Among them, ρ k It is the energy consumption coefficient related to UAM aircraft, which depends on the CPU performance structure.

7. The UAM-assisted MEC cache system based on federated DDQN according to claim 5 is characterized in that: When the UAM aircraft applies for content services from the drone, there are two situations. The first is when the drone has cached the required content services locally, and the drone directly calculates the UAM aircraft task. The second is when the drone does not have the required services cached, the drone applies to the macro base station, and the macro base station processes the task.

8. The UAM-assisted MEC cache system based on federated DDQN according to claim 5, characterized in that: When the local area directly requests service from the macro base station, the transmission delay when the UAM aircraft k transmits the task to the macro base station for: The device assigns the task to the macro base station to calculate the delay T b for: Among them, r b，k Indicates the transmission rate between the UAM aircraft and the macro base station; F b represents the average computing power of the macro base station; C k Indicates the CPU cycles required for the task; L k Indicates the size of the task.

Images