CN111629443B - Optimization method and system for dynamic spectrum slicing frame in super 5G Internet of vehicles - Google Patents
Optimization method and system for dynamic spectrum slicing frame in super 5G Internet of vehicles Download PDFInfo
- Publication number
- CN111629443B CN111629443B CN202010523949.5A CN202010523949A CN111629443B CN 111629443 B CN111629443 B CN 111629443B CN 202010523949 A CN202010523949 A CN 202010523949A CN 111629443 B CN111629443 B CN 111629443B
- Authority
- CN
- China
- Prior art keywords
- service
- resource
- time slot
- unmanned aerial
- optimization
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W72/00—Local resource management
- H04W72/04—Wireless resource allocation
- H04W72/044—Wireless resource allocation based on the type of the allocated resource
- H04W72/0453—Resources in frequency domain, e.g. a carrier in FDMA
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
Abstract
The invention discloses an optimization method of a dynamic spectrum slice frame in a super 5G Internet of vehicles, which comprises the following steps: for a particular type of service request, scheduling service requests arriving within the time slot into different downlink resource planes to deliver the service; maintaining a queue for each type of service request by each resource plane in a resource slicing mode; determining the number of unmanned aerial vehicles to be scheduled in each time slot to meet the dynamic packet processing requirements from all service requests according to the unmanned aerial vehicle packet processing capacity of each resource plane in each time slot; packet processing capabilities for each service request within a time slot according to each resource plane; and decoupling optimization parameters based on the Lyapunov optimization technology to determine optimal control parameters by taking the minimization of service supply cost and the maximization of service utility of all resource planes as optimization targets. The invention effectively improves the service performance and the resource utilization rate of the user.
Description
Technical Field
The invention relates to the technical field of super-5G vehicle networking, in particular to an optimization method and system for a dynamic spectrum slice frame in the super-5G vehicle networking.
Background
To achieve extensive on-board service, modern interconnected vehicles require a robust network that simultaneously guarantees ultra-reliable and low-latency communications for assisted/autonomous driving, provides high-bandwidth on-board entertainment services, supports communication requirements for intensive task offloading, etc. Since different applications share the same spectrum resources and their traffic consumption patterns are different, they have significant impact on each other, which makes it difficult to achieve satisfactory quality of service. For example, high bandwidth services are highly prone to channel blockage, resulting in packet loss or transmission delay for lightweight, ultra-reliable and low latency services. Recently, network virtualization and slicing technology is considered as an indispensable component in next generation networks, and network slicing refers to running multiple private networks on a shared physical network infrastructure, wherein slices are completely isolated from each other, and each slice can customize service-oriented functions or can be combined to meet different service requirements.
On the other hand, the air-ground integrated internet of vehicles (i.e., integrated satellite network, drone network, and terrestrial cellular network) is a promising architecture of the super 5G internet of vehicles (B5G, 6G-oriented internet of vehicles with the introduction of drone and satellite network). In the air-ground integrated internet of vehicles, besides the ground cellular network, there are other heterogeneous networks, in which a satellite network can provide ubiquitous wireless coverage, and an unmanned aerial vehicle network can assist the ground network in improving its flexibility. However, implementing efficient slicing functions in such networks is a difficult task. Firstly, the space-air-ground integrated Internet of vehicles spectrum resources are multidimensional and heterogeneous, and access technologies of the space-air-ground integrated Internet of vehicles spectrum resources are different, so that the space-air-ground integrated Internet of vehicles spectrum resources are difficult to manage and slice; secondly, the network topology changes violently due to the rapid movement of the vehicle, so that the service request is unpredictable; third, how to achieve a balance between system performance and drone scheduling cost is also crucial.
The current technology has limited research on the network slice of the air-space-ground integrated Internet of vehicles. There are some related researches to roughly discuss the car networking slicing technology, but no specific implementation details are given, while other researches are to perform network slicing in different scenes and mainly focus on content pushing and caching slicing resources. However, the vehicle-mounted network environment is more widely existed in the real environment, the spectrum resource slicing is more complex, and the research on the spectrum resource slicing and the management optimization technology for the air-space-ground integrated vehicle networking has better application prospect and use value.
Disclosure of Invention
The invention provides an optimization method and system for a dynamic spectrum slicing frame in a super-5G Internet of vehicles, which are used for solving the technical problem that efficient spectrum slicing is difficult to realize in an air-space-ground integrated Internet of vehicles.
In order to solve the technical problems, the technical scheme provided by the invention is as follows:
an optimization method for a dynamic spectrum slicing frame in a super 5G Internet of vehicles comprises the following steps:
for a particular type of service request, scheduling service requests arriving within a time slot into different downlink resource planes to deliver the service; maintaining a queue for each type of service request by each resource plane in a resource slicing mode;
determining the number of unmanned aerial vehicles to be scheduled in each time slot to meet the dynamic packet processing requirements from all service requests according to the unmanned aerial vehicle packet processing capacity of each resource plane in each time slot;
packet processing capability for each service request within a time slot according to each resource plane; the method comprises the following steps of taking minimization of service supply cost and maximization of service effectiveness of all resource planes as optimization targets, decoupling optimization parameters based on a Lyapunov optimization technology to determine optimal control parameters, wherein the control parameters comprise: request quantity a of ith service admittance in time slot t i (t) mixing a i (t) scheduling of the requested amount to a requested amount a in a jth downlink resource plane ij (t) dynamically scheduling the processing capacity C (t) provided by the UAV, and the packet processing capacity b of the jth resource plane for the ith service request in the time slot t ij (t)。
As a further improvement of the process of the invention:
preferably, the control parameters satisfy the following constraints:
a i (t)=∑ j∈J a ij (t)
C(t)∈[0,C max ]
wherein, I ═ 1, 2., I } denotes a set of services, I ∈ I, J ∈ {0, 1., J } denotes a set of downlink resources, a downlink resource plane T ═ 0,1, 2., T, } denotes a timeslot in which the network operates, and T is a timeslot; z is a linear or branched member j (t) (J ∈ J) is the packet processing capacity of the jth resource plane in the tth time slot; c max The number of unmanned aerial vehicles which can be actually dispatched; b max Is the maximum number of requests processed in a slot; a. the i (t) denotes a certain type of service volume arriving within a time slot t, and A i (t) are independently co-distributed in different time slots and independent of other arriving service types.
Preferably, the control parameters satisfy the following criteria:
queue length Q for ith service request in time slot t ij (t) having:
Q ij (t+1)=max[Q ij (t)-b ij (t),0]+a ij (t), (3)
Satisfies the following conditions:
time-averaged queue stability criterion:
for the drone resource plane, its packet processing capability depends on the real-time scheduling of the drones, i.e.:
wherein, Z j (t) where j-0 represents the drone resource plane, Z 0 (t) indicates available drone packet processing capacity within time slot t; b is a mixture of i0 (t) packet processing capability for the unmanned aerial vehicle resource plane for the ith service request within time slot t;
long term time averaged equivalence:
where θ is the desired unmanned aerial vehicle processing capability across time slices, Z 0 (0) Can be initialized to any non-negative value.
Preferably, with the minimization of service provisioning cost and maximization of service utility for all resource planes as optimization objectives, the optimization problem is as follows:
wherein the system utility function:
wherein λ i A positive real number representing the ith service weight,is the long-term time-averaged admission request volume for the ith service; long term average dispatch cost of droneBeta is a unit cost representing the processing capacity of the unmanned aerial vehicle and is a weight constant.
Preferably, with the minimization of service supply cost and maximization of service utility for all resource planes as optimization objectives, the optimization problem is as follows:
wherein the system utility function:
wherein λ i A positive real number indicating the ith service weight,is the long-term time average admission request volume for the ith service; long term average dispatch cost of droneBeta is a unit cost representing the processing capacity of the unmanned aerial vehicle, and is a weight constant, gamma i (t) is a i (t) non-negative auxiliary variables.
Preferably, when decoupling optimization parameters based on the lyapunov optimization technology, a lyapunov drift plus penalty minimization algorithm is adopted, and the conditional lyapunov drift plus penalty of a single time slot is as follows:
whereinIs the conditional lyapunov offset for a single slot, and V is a weight constant, used to trade off between system gain and queue stability; delta of V (t) indicates more stable queue length and higher system revenue;a connection vector representing the instantaneous queue size at the end of the t-th slot for all queues; queueThe virtual queues in the set of (a) are:
whereinIs a set of the lengths of the queues,is the set of available drone packet processing capabilities within the time slot t.
Preferably, Δ V The upper limit of (t) is:
wherein B is defined by the upper limit A max ,b max ,C max Determined by max Admission to all servicesLimit, b max Is an upper bound on the processing of requests within a slot, C max Is the upper limit of the unmanned aerial vehicle which can be dispatched practically.
Preferably, at each time slot t, the status Θ (t) of all queues is observed based on the following five control phases;
Resource slicing: for each resource plane, sequencing the queues according to the descending order of the weights of the resource planes, and segmenting resources in sequence as long as a resource constraint formula (2) is met;
and (3) queue updating: queue state Q after all control parameters are obtained ij (t),H i (t) and Z 0 (t) is updated based on equations (3), (11) and (5), respectively, for making a decision in the next time slot.
The invention also provides a computer system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of any of the above methods when executing the computer program.
The invention has the following beneficial effects:
1. the optimization method and the optimization system for the dynamic spectrum slicing frame in the super-5G Internet of vehicles perform dynamic slicing on resources of the super-5G Internet of vehicles (aerospace-geostationary integrated vehicle-mounted network) and manage and optimize the spectrum resources of the super-5G Internet of vehicles so as to realize isolation of service resource supply. In order to provide service isolation, the system can independently establish a service queue for each type of network service and supply spectrum resources, the dynamic spectrum slice framework comprises user request control, request scheduling, unmanned plane scheduling and resource slicing, and the optimization aim is to maximize the long-term benefit of the network system under the condition that the service queue is stable. Based on the Lyapunov optimization technology, the dynamic slice optimization problem is decoupled into a plurality of independent sub-problems and solved. A large number of simulation results show that the method has good effects in the aspects of improving the system throughput, efficiently using the unmanned aerial vehicle, balancing contradictions between the system benefit and the stability and the like.
2. In a preferred scheme, the optimization method and the optimization system for the dynamic spectrum slice frame in the super-5G vehicle networking keep the stability of the queue, can flexibly balance the stability and the benefits of the system, realize the balance between the throughput of the system and the unmanned aerial vehicle by adjusting control parameters, and effectively improve the service performance of users and the resource utilization rate.
In addition to the objects, features and advantages described above, other objects, features and advantages of the present invention are also provided. The present invention will be described in further detail below with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification. In the drawings:
FIG. 1 is a system model diagram of an embodiment of the present invention;
fig. 2 is a schematic diagram of a dynamic slicing workflow according to an embodiment of the present invention, including: request control, request scheduling, unmanned aerial vehicle scheduling and resource slicing;
FIG. 3 is a schematic diagram of a simulation experiment scenario according to an embodiment of the present invention;
FIG. 4 is a diagram illustrating the size and stability of a simulation service queue according to an embodiment of the present invention;
FIG. 5 is a schematic diagram illustrating a length of a time-averaged service queue under different values of the weight coefficient V in a simulation experiment according to an embodiment of the present invention;
fig. 6 is a schematic diagram of time-averaged unmanned aerial vehicle resource conditions under different values of weight coefficient V in a simulation experiment according to an embodiment of the present invention;
FIG. 7 is a schematic diagram illustrating a time-averaged profit variation under different values of the weight coefficient V in a simulation experiment according to an embodiment of the present invention;
FIG. 8 is a schematic diagram illustrating a variation of a queue length averaged over time in a simulation experiment according to an embodiment of the present invention under different values of the weighting factor V;
FIG. 9 is a schematic diagram of the time-averaged throughput of the system in the simulation experiment according to the embodiment of the present invention;
fig. 10 is a schematic diagram of the time-averaged queue length in the simulation experiment according to the embodiment of the present invention.
Detailed Description
The embodiments of the invention will be described in detail below with reference to the drawings, but the invention can be implemented in many different ways as defined and covered by the claims.
Fig. 1 is a typical model of the air-space-ground integrated internet of vehicles system in this embodiment, in which a fixed ground network is formed at a ground base station to provide cellular network connection, highly flexible drones are dynamically scheduled to provide an on-demand aeronautical network, and an orbiting satellite synchronized with the equator is provided to serve a complete wireless coverage area of interest. Assuming that each vehicle has three network interfaces, namely a cellular network, an unmanned aerial vehicle and satellite communication, since the vehicle network service mainly depends on the performance of a downlink, the downlink service scenario of the air-ground integrated vehicle networking is mainly considered in the embodiment.
On the basis of the system model, the invention adopts a dynamic spectrum slicing frame to dynamically slice the air-space-ground integrated vehicle-mounted network resources and manage and optimize the spectrum resources so as to realize the isolation of service resource supply. By software and slicing of spectrum resources, requests for each type of service can be processed and serviced in a separate queue. The dynamic spectrum slicing framework of the invention considers that online decision is made under each time slot, and comprises the following contents:
(a) controlling how many different service requests the system should access;
(b) how to schedule resources of the satellite, drone and terrestrial cellular networks to process for each type of request;
(c) if needed, determining how many drones should be scheduled to supplement spectrum resources;
(d) how to slice the spectrum to reduce the queue backlog for the service queues on each resource plane;
moreover, the invention mainly considers three core requirements of the vehicle network system:
(e) the system should access and process as many service requests as possible;
(f) after the system performance is ensured, the scheduling cost of the unmanned aerial vehicle is reduced as much as possible;
(g) the backlog of queues should be stabilized within a certain range for each type of service.
Under the above decision and demand conditions, the optimization method for the dynamic spectrum slice frame in the super 5G vehicle networking of the invention comprises the following steps:
for a particular type of service request, scheduling service requests arriving within the time slot into different downlink resource planes to deliver the service; maintaining a queue for each type of service request by each resource plane in a resource slicing mode;
determining the number of unmanned aerial vehicles to be scheduled in each time slot to meet the dynamic packet processing requirements from all service requests according to the unmanned aerial vehicle packet processing capacity of each resource plane in each time slot;
packet processing capabilities for each service request within a time slot according to each resource plane; decoupling optimization parameters based on a Lyapunov optimization technology to determine optimal control parameters by taking minimization of service supply cost and maximization of service utility of all resource planes as optimization targets, wherein the control parameters comprise: request quantity a of ith service admission in time slot t i (t) mixing a i (t) scheduling of the requested amount into the requested amount a in the jth downlink resource plane ij (t), dynamically scheduling drone offeringsAnd the processing capacity b of the jth resource plane for the packet of the ith service request in the time slot t ij (t)。
In practical applications, the present invention can be optimized based on the above steps, which are exemplified as follows:
the definition I ═ {1, 2., I } denotes a set of services, and J ═ 0, 1., J } denotes a set of downlink resources. The network operates in time slots T ═ 0,1, 2. Fig. 2 shows a workflow of dynamic slicing in this embodiment, which includes the following steps:
s1, request control: for a certain type of service I ∈ I, A i (t) represents the amount of service arrived in time slot t, assume A i (t) are independently and identically distributed in different time slots and independent of other arriving service types. Not all requests can arrive due to the limited downstream capacity of the entire network. Definition a i (t) is the amount of requests for the ith service admission in time slot t, thus:
s2, request scheduling: next, the admitted requests need to be scheduled into different downlinks to deliver the service. a is ij (t) represents a i (t) scheduling the requested amount into a jth downlink resource plane. Then there is a i (t)=∑ j∈J a ij (t) of (d). By way of resource slicing, each resource plane maintains a dedicated queue for each type of service request, as shown in FIG. 1.
S3, unmanned aerial vehicle scheduling: definition of Z j (t) (J ∈ J) is the packet processing capacity of the jth resource plane in the tth slot. Consider that in a high-density vehicle scenario, where network resources are insufficient to meet a busy service arrival, drone dispatch is used on demand. Denote the drone resource plane by j ═ 0, Z 0 (t) indicates the available drone packet processing capacity in time slot t, and c (t) indicates the processing capacity provided by dynamically scheduling drones. C (t) since the number of schedulable drones is limited in practical cases) Satisfying the bounded hypothesis C (t) e [0, C max ]. Within each time slot t, the system needs to determine the number of drones scheduled to meet the dynamic packet processing requirements from all services.
S4, resource slicing: definition b ij (t) packet processing capability for the ith service request in time slot t for the jth resource plane. Thus:
in the present embodiment, only a limited number of requests can be processed in one time slot for each type of service request, i.e., the following bounded assumption holds,
to perform the above four steps, it is necessary to determine the following control parameters, i.e., a, in each time slot i (t),a ij (t),C(t),b ij (t) of (d). In order to make these control decisions, it is first necessary to describe the impact of these parameters on the system performance. With Q ij (t) indicates the queue length of the ith service request in time slot t, and there are:
Q ij (t+1)=max[Q ij (t)-b ij (t),0]+a ij (t), (3)
whereinIn order to achieve a feasible iteration, the stability of all queues must be guaranteed, so a time-averaged queue stability criterion is adopted, defined as follows:
For the drone resource plane, its packet processing capability depends on the real-time scheduling of the drone, i.e.:
in the embodiment, the time delay of the scheduling and operation of the unmanned aerial vehicle is considered, and in practical application, the available processing capacity of the unmanned aerial vehicle should be kept at a certain level to ensure the continuity of the service. It can translate into the following long-term time-averaged equivalents:
where θ is the desired unmanned aerial vehicle processing capability across time slices, Z 0 (0) Can be initialized to any non-negative value.
Problem construction:
in this embodiment, different services typically have different quality of service, which also provides different levels of utility to the network service provider. To achieve fair service provisioning, the following system utility functions are defined:
wherein λ i A positive real number representing the ith service weight,is the long-term time-averaged admission request volume for the ith service. In particular, due to utility functionAre all convex functions, and equation (7) ensures diminishing returnsI.e. the service provider cannot guarantee to continuously increase its overall utility by only increasing admission requests for services.
At the same time, in order to maximize revenue, the service provisioning cost for all resource planes should be minimized. Since the processing power of cellular networks and satellites is static, network service providers need to minimize the cost of drone dispatch. Defining the long-term average dispatch cost of the unmanned plane asIn fact, it is difficult to simultaneously compromise drone dispatch cost minimization and service utility maximization. Thus, in the next step, a weighted sum of the two objectives will be taken and form the following optimization problem:
where β is a weight constant, and refers to the unit cost of scheduling the processing power of the drones.
In this embodiment, it is challenging to directly solve the above optimization problem for the following reasons, first, the number of request arrivals for different services is time-varying and a priori unknown, so it is difficult to satisfy the offline optimal decision of the constraint condition (1); second, since the processing power of the drone dispatched from the t slot is only available from the t +1 slot, a set of decision parameters that satisfy constraint (2) may not be able to serve all admission requests in the next slot, resulting in potential request accumulation and long-term queue instability. Therefore, there is a continuing need to use the lyapunov optimization framework to decouple the correlation between the optimization parameters and provide efficient control decisions across each slot of a time slice, collectively solving the optimization problem described above.
Problem transformation: lyapunov drift plus penalty minimization algorithm in optimizing long termThe inter-average target aspect is very effective. Applying the Qinsheng inequality to a set of convex effects functions, problemThe transformation may be done by introducing a set of auxiliary variables. Definition of gamma i (t) is a i (t), then the following reasoning holds:
in this example, the problem after transformationFalling into the category of the lyapunov optimization framework can be solved by constructing a virtual queue corresponding to the auxiliary variable constraint in (10). Specifically, for each service type, a virtual queue is defined:
as long as the queueIs stable, the constraint (10) can be satisfied. Also to further illustrate the constraint of resource slicing in condition (2), considerA high traffic scenario is assumed, i.e. the processing power of the satellite and the cellular network is always fully utilized. Then in each time slot, there areAnd is provided with
Consider a queueAndover a time slice, whereinIs a set of the lengths of the queues,is the set of available drone packet processing capabilities within the time slot t. Problem to be solvedCan be defined as:
the above definition represents one scalar measure of all queue sizes, typically non-negative, and l (t) is 0 if and only if all queue sizes are 0. Order toA connection vector representing the instantaneous queue size at the end of the t-th slot for all queues. The conditional lyapunov drift plus penalty for a single slot is defined as:
whereinIs the conditional lyapunov offset for a single slot, and V is a weight constant to trade off the trade-off between system gain and queue stability. A smaller Δ V (t) indicates a more stable queue length and higher system revenue. Delta of V The upper limit of (t) can be derived from the following lemma.
2, leading: for all possible queue states and control actions, Δ V The upper limit of (t) is:
wherein B is defined by the upper limit A max ,b max ,C max Determined by max Upper bound for admission of all services, b max Is an upper limit for processing requests within a time slot, C max Is the upper limit of the unmanned aerial vehicle which can be dispatched practically.
Decoupling of the control parameters to the right of the drift plus penalty upper bound (14) enables us to align gamma i (t),a i (t),a ij (t),C(t),b ij (t) making independent and sequential decisions. Next, an attempt is made to minimize four independent conditional terms in the boundary (14).
Dynamic slicing algorithm: at each time slot t, first, the states Θ (t) of all queues are observed based on the following five control phases.
P1. auxiliary variable decision: i.e. state H of a given observation queue i (t), minimization. The optimization problem is a feasible set gamma for solving three univariate optimization problems i And they all use a function effectivelyClosed-form solution of (c).
P2. request control and scheduling: given queue State H i (t) and Q ij (t) of (d). Due to coupling constraintsDetermining a group a i (t) and a ij (t) to maximize per serviceIt is difficult, therefore, to use a connection-based shortest queue first heuristic to determine a sequentially i (t) and a ij (t) of (d). For the ith type of service, a simple scheduling policy is to put all allowed requests a i (t) sending to the resource plane with the shortest queue, i.e. a ij (t)=a i (t) of (d). If it is notThen a ij (t) is 0. Then, it reduces to maximumAt a i And (t) is easier to solve in a feasible interval.
And P3, unmanned plane scheduling: based on instantaneous queue size Z 0 (t), predefined parameters β, θ, V and the feasible interval C (t) E [0, C max ]. By minimizingThe optimal unmanned aerial vehicle dispatch capacity can be easily found.
P4. resource slicing: to maximize the last term in equation (14), it is sufficient to make a separate allocation decision on each resource plane. For the satellite and cellular planes (j ∈ {1,2}), give b ij (t) assigned a weight of Q ij (t)+Z j And the weight assigned to the drone resource plane (j ═ 0) is Q i0 (t)+Z 0 (t) - θ. An effective solution is to, for each resource plane, first order the queues according to the order in which their weights decrease,the resources are then partitioned in order as long as the resource constraint (2) is satisfied. This stage does not consider queues with negative weights, and b is the corresponding ij (t) is set to 0.
P5. queue update: queue State Q after obtaining all control parameters ij (t),H i (t) and Z 0 (t) can be used to make a decision in the next slot based on the (3), (11) and (5) updates, respectively.
In summary, the embodiment adopts the lyapunov optimization technology to maximize the long-term benefit of the system. Specifically, firstly, the time-averaged queue backlog of all services is considered, a quadratic Lyapunov function is constructed, and a system yield function combining the system average throughput and the unmanned aerial vehicle scheduling cost is provided. In order to maximize the system gain, a reverse penalty function is also constructed. Furthermore, in order to stabilize the system while minimizing the time-averaged penalty, the drift plus penalty should be limited to a minimum. And decomposing the control problem into deterministic subproblems such as auxiliary variable setting, request control, request scheduling, unmanned aerial vehicle scheduling, resource slicing and the like based on a derived drift plus penalty theory.
Simulation experiment:
the technical scheme of the invention is demonstrated by simulation experiments.
As shown in fig. 3, the simulation experiment defines a bidirectional 8-lane highway scene, and uses the simulation tool SUMO to generate a real vehicle track. The entrance of each passage generated by a vehicle and the arrival process of the vehicle are considered to be a poisson distribution lambda V Of arrival, λ V 0.25. There is a cellular base station in the center of the road with coordinates (R2000 m, D100 m, H50 m), and if a drone is derived, it hovers at coordinates (x, 0m, 5 m) (e.g., H5 m).
Cellular parameters: respectively set total bandwidth W B Is 20MHz, transmission power p B 20W, background noise density n 0 Is 1 x 10 - 15 W/Hz, and a gain of channel power ofWherein sigma Bi Is the belief fading and is assumed to follow an exponential distribution of unity mean.Is the distance between the vehicle and the base station, α is the path loss coefficient and α is 3. In addition, there are 100 orthogonal subchannels K, which may be allocated to users, and the bandwidth of each subchannel is 0.2 MHz. Due to the data rate of the sub-channels being subject to distanceIn the long term, the data rate of a subchannel can be expressed as Where y is the distance the vehicle travels from road entry point a to road center point B, so the initial transport capacity of the cellular resource plane can be obtained.
Unmanned aerial vehicle communication parameters: the method comprises the steps of adopting a WiFi technology to achieve unmanned aerial vehicle communication, and adopting a drive-thru area-based model to achieve unmanned aerial vehicle throughput acquisition, wherein continuous communication areas exist, and each area has different data rates. In particular, an 802.11n protocol is employed, where the bandwidth operates at 2.4GHz and the total bandwidth is set to 40 MHz. The length of each region and the associated data rate are calculated under a free space path loss model according to the WiFi data table.
Satellite communication parameters: the attenuation of the satellite-link is modeled based on the Weibull channel model, considering that the vehicle can always use the satellite connection. Specifically, assuming 20 subchannels, each subchannel providing a data rate of 0.5Mbps, the transmission capacity of the satellite resource plane is therefore
Vehicle service activities: in the above-described dicing frame, three typical vehicles are consideredA network service. Specifically, an ultra-reliable low-latency service is initiated for each vehicle, where each vehicle downloads one packet per time slot. In order to simulate automatic high-precision map downloading service or assist calculation-intensive task unloading, the embodiment provides a streaming media service for each vehicle, and requires to maintain a relatively stable throughput T during network connection strm (set to 0.5 Mbps). For the vehicle network, the user may randomly initiate a download service. In particular, a user may initiate a download request during a Poisson process at a rate of 0.2, and the amount of download may be from the set {10 } 5 ,10 6 ,...,10 10 Randomly selected.
Fig. 4 shows the stability and size (in units of packets) of the service queues of three vehicle-mounted applications in the simulation experiment of the present invention, where the queue of each application service is very stable and the queue length is positively correlated to the request requirement.
Fig. 5 shows the average time length of the service queue under different values of the weighting factor V in the simulation experiment of the present invention. The average queue may also stabilize well when the request is far beyond the processing power of the system. When the time slot reaches 4000, the increase in the time-averaged queue size tends to be smooth. Furthermore, the time-averaged queue increases as V increases. (larger V indicates that the system is more concerned with revenue rather than queue stability)
Fig. 6 shows the time-averaged resource condition of the unmanned aerial vehicle under different values of the weight coefficient V in the simulation experiment of the present invention. After initial adjustment, the unmanned aerial vehicle resources can be well stabilized by scheduling the unmanned aerial vehicle, but the larger V results in lower stability of the unmanned aerial vehicle resources.
FIG. 7 shows the time-averaged profit variation under different values of the weighting factor V in the simulation experiment of the present invention. As the value of V increases, the time-averaged benefit increases significantly, but to satisfy the stability constraint, the benefit is bounded above regardless of how the value of V is increased. Specifically, when V ranges from 0 to 106, the time-averaged benefit increases to 23:4, and even if V is too high (e.g., 1010), the benefit is around 23: 4.
FIG. 8 shows a simulation experiment of the present invention withoutThe time-averaged queue size varies for the same value of V. As the value of V increases, the average queue size also increases. However, when the value of V is from 5X 10 5 Becomes 5.5X 10 5 There is a sudden rise. As can be seen from FIG. 5, the V value is less than 5X 10 5 Meanwhile, the resource cost of the unmanned aerial vehicle is reduced, and when the V value is 5.5 multiplied by 10 5 No additional drone resources are provisioned, resulting in a sudden increase in time-averaged revenue and queue size.
Fig. 9 and 10 show the time-averaged throughput and the time-averaged queue size of the system in the simulation experiment of the present invention. The following conclusions can be drawn from the figure: first, compared with the dynamic slicing scheme, since the system throughput is rapidly decreased and the queue size is rapidly increased, the fixed slicing scheme cannot keep up with the vehicle state, and the resource management efficiency is low. For example, a fixed slice provides a throughput of approximately 1750 packets per slot, while a dynamic slice can increase this to 2200 packets/slot, an improvement of approximately 26%. Secondly, the upgraded dynamic slicing scheme can achieve the same level of throughput performance and queue stability as dynamic slicing. Therefore, the proposed dynamic slicing not only can improve the overall service performance of the system, but also can support differentiated service provisioning. In addition, by the upgraded dynamic slice, the communication service with ultra-high reliability and low delay can be ensured, and the reduction of the throughput performance can be ignored.
The present invention also provides a computer system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the steps of any of the above embodiments being implemented when the computer program is executed by the processor.
In conclusion, the method and the device keep the stability of the queue, can flexibly balance the stability and the benefits of the system, realize the balance between the throughput of the system and the unmanned aerial vehicle by adjusting the control parameters, and effectively improve the service performance and the resource utilization rate of the user. A large number of simulation results show that the method has good effects in the aspects of improving the system throughput, efficiently using the unmanned aerial vehicle, balancing contradictions between the system benefit and the stability and the like.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (4)
1. An optimization method for a dynamic spectrum slicing frame in a super 5G Internet of vehicles is characterized by comprising the following steps:
for a particular type of service request, scheduling service requests arriving within the time slot into different downlink resource planes to deliver the service; maintaining a queue for each type of service request by each resource plane in a resource slicing mode;
determining the number of unmanned aerial vehicles to be scheduled in each time slot to meet the dynamic packet processing requirements from all service requests according to the unmanned aerial vehicle packet processing capacity of each resource plane in each time slot;
packet processing capability for each service request within a time slot according to each resource plane; decoupling optimization parameters based on Lyapunov optimization technology with the service supply cost minimization and the service utility maximization of all resource planes as optimization targets to determine optimal control parameters, wherein the control parameters comprise: request quantity a of ith service admittance in time slot t i (t) mixing a i (t) scheduling of the requested amount into the requested amount a in the jth downlink resource plane ij (t) dynamically scheduling the processing capacity C (t) provided by the drone and the packet processing capacity b of the jth resource plane for the ith service request in the time slot t ij (t);
The control parameters satisfy the following constraints:
a i (t)=∑ j∈J a ij (t)
C(t)∈[0,C max ]
wherein, I ═ 1, 2., I } denotes a set of services, I ∈ I, J ∈ {0, 1., J } denotes a set of downlink resources, a downlink resource plane T ═ 0,1, 2., T, } denotes a timeslot in which the network operates, and T is a timeslot; z j (t) (J ∈ J) is the packet processing capacity of the jth resource plane in the tth time slot; c max The number of unmanned aerial vehicles which can be actually dispatched; b max Is an upper limit for processing requests within a slot; a. the i (t) represents a certain type of service volume arriving within the time slot t, and A i (t) independently co-distributed in different timeslots and independent of other arriving service types;
the control parameters meet the following criteria:
queue length Q for ith service request at time slot t ij (t) having:
Q ij (t+1)=max[Q ij (t)-b ij (t),0*+a ij (t), (3)
Satisfies the following conditions:
time-averaged queue stability criterion:
for the drone resource plane, its packet processing capability depends on the real-time scheduling of the drone, i.e.:
wherein, Z j (t) where j is 0 denotes the drone resource plane, Z 0 (t) indicates available drone packet processing capacity within time slot t; b is a mixture of i0 (t) packet processing capability for the unmanned aerial vehicle resource plane for the ith service request within time slot t;
long term time averaged equivalence:
where θ is the desired unmanned aerial vehicle processing capability across time slices, Z 0 (0) Can be initialized to any non-negative value;
with the optimization objectives of minimizing the service supply cost and maximizing the service utility of all resource planes, the optimization problem is as follows:
wherein the system utility function:
wherein λ i A positive real number indicating the ith service weight,is the long-term time average admission request volume for the ith service; the long-term average scheduling cost of the unmanned aerial vehicle isBeta is unit cost representing the processing capacity of the unmanned aerial vehicle, and is a weight constant;
with the optimization objectives of minimizing the service supply cost and maximizing the service utility of all resource planes, the optimization problem is as follows:
wherein the system utility function:
wherein λ i A positive real number representing the ith service weight,is the long-term time average admission request volume for the ith service; the long-term average scheduling cost of the unmanned aerial vehicle isBeta is a unit cost representing the processing capacity of the unmanned aerial vehicle, and is a weight constant, gamma i (t) is a i (t) a non-negative auxiliary variable;
when decoupling optimization parameters based on the Lyapunov optimization technology, a Lyapunov drift plus penalty minimization algorithm is adopted, and the conditional Lyapunov drift plus penalty of a single time slot is as follows:
whereinIs the conditional lyapunov offset for a single slot, and V is a weight constant, used to trade off between system revenue and queue stability; delta V (t) indicates more stable queue length and higher system revenue;a connection vector representing the instantaneous queue size of all queues at the end of the tth slot; queueThe virtual queues in the set are:
2. The optimization method for dynamic spectrum slicing framework in ultra-5G internet of vehicles according to claim 1, wherein the Δ is V The upper limit of (t) is:
wherein B is defined by the upper limit A max ,b max ,C max Determined by max Upper bound for admission of all services, b max Is an upper limit for processing requests within a time slot, C max Is the upper limit of the unmanned aerial vehicle which can be dispatched practically.
3. The optimization method for the dynamic spectrum slicing frame in the ultra-5G vehicle networking according to the claim 1, wherein in each time slot t, the states Θ (t) of all queues are observed based on the following five control phases;
Resource slicing: for each resource plane, sequencing the queues according to the descending order of the weights of the resource planes, and segmenting resources in sequence as long as a resource constraint formula (2) is met;
and (3) queue updating: queue state Q after all control parameters are obtained ij (t),H i (t) and Z 0 (t) is updated based on equations (3), (11) and (5), respectively, for making a decision in the next time slot.
4. A computer system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method of any of claims 1 to 3 when executing the computer program.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010523949.5A CN111629443B (en) | 2020-06-10 | 2020-06-10 | Optimization method and system for dynamic spectrum slicing frame in super 5G Internet of vehicles |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010523949.5A CN111629443B (en) | 2020-06-10 | 2020-06-10 | Optimization method and system for dynamic spectrum slicing frame in super 5G Internet of vehicles |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111629443A CN111629443A (en) | 2020-09-04 |
CN111629443B true CN111629443B (en) | 2022-07-26 |
Family
ID=72261065
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010523949.5A Active CN111629443B (en) | 2020-06-10 | 2020-06-10 | Optimization method and system for dynamic spectrum slicing frame in super 5G Internet of vehicles |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111629443B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112565377B (en) * | 2020-11-30 | 2021-09-21 | 北京邮电大学 | Content grading optimization caching method for user service experience in Internet of vehicles |
CN112866368B (en) * | 2021-01-12 | 2022-03-18 | 北京邮电大学 | Air-ground remote Internet of things design method and system |
CN113296909B (en) * | 2021-05-14 | 2022-11-11 | 上海交通大学 | Optimized service deployment method in mobile edge computing |
CN117082626B (en) * | 2023-07-26 | 2024-03-26 | 北京交通大学 | Service multiplexing slice resource allocation method and system for train wireless communication network |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107682135A (en) * | 2017-09-30 | 2018-02-09 | 重庆邮电大学 | A kind of network slice adaptive virtual resource allocation method based on NOMA |
EP3297227A1 (en) * | 2016-09-14 | 2018-03-21 | Deutsche Telekom AG | Method for routing in a communication network, communication network, program and computer program product |
CN108965024A (en) * | 2018-08-01 | 2018-12-07 | 重庆邮电大学 | A kind of virtual network function dispatching method of the 5G network slice based on prediction |
CN110366193A (en) * | 2019-05-28 | 2019-10-22 | 中国电力科学研究院有限公司 | A kind of splitting and disposing method and device of the network arrangements layer carrying of smart grid |
CN110809261A (en) * | 2019-10-31 | 2020-02-18 | 重庆邮电大学 | Network slice dynamic resource scheduling method for joint congestion control and resource allocation in H-CRAN network |
CN111258677A (en) * | 2020-01-16 | 2020-06-09 | 重庆邮电大学 | Task unloading method for heterogeneous network edge computing |
-
2020
- 2020-06-10 CN CN202010523949.5A patent/CN111629443B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3297227A1 (en) * | 2016-09-14 | 2018-03-21 | Deutsche Telekom AG | Method for routing in a communication network, communication network, program and computer program product |
CN107682135A (en) * | 2017-09-30 | 2018-02-09 | 重庆邮电大学 | A kind of network slice adaptive virtual resource allocation method based on NOMA |
CN108965024A (en) * | 2018-08-01 | 2018-12-07 | 重庆邮电大学 | A kind of virtual network function dispatching method of the 5G network slice based on prediction |
CN110366193A (en) * | 2019-05-28 | 2019-10-22 | 中国电力科学研究院有限公司 | A kind of splitting and disposing method and device of the network arrangements layer carrying of smart grid |
CN110809261A (en) * | 2019-10-31 | 2020-02-18 | 重庆邮电大学 | Network slice dynamic resource scheduling method for joint congestion control and resource allocation in H-CRAN network |
CN111258677A (en) * | 2020-01-16 | 2020-06-09 | 重庆邮电大学 | Task unloading method for heterogeneous network edge computing |
Non-Patent Citations (2)
Title |
---|
5G网络端到端切片技术研究;乔楚;《通信技术》;20180910(第09期);全文 * |
基于QoS的5G切片间资源分配;王文琪等;《光通信研究》;20180529(第03期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN111629443A (en) | 2020-09-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111629443B (en) | Optimization method and system for dynamic spectrum slicing frame in super 5G Internet of vehicles | |
Liu et al. | Deep reinforcement learning based latency minimization for mobile edge computing with virtualization in maritime UAV communication network | |
Lyu et al. | Service-oriented dynamic resource slicing and optimization for space-air-ground integrated vehicular networks | |
CN113346944B (en) | Time delay minimization calculation task unloading method and system in air-space-ground integrated network | |
CN111552313B (en) | Multi-unmanned aerial vehicle path planning method based on edge calculation dynamic task arrival | |
Yang et al. | Online trajectory and resource optimization for stochastic UAV-enabled MEC systems | |
CN110602633B (en) | Explosive flow-oriented mobile edge computing unmanned aerial vehicle cluster auxiliary communication method | |
US11567512B1 (en) | Unmanned aerial vehicle (UAV) task cooperation method based on overlapping coalition formation (OCF) game | |
Sun et al. | Enhancing the user experience in vehicular edge computing networks: An adaptive resource allocation approach | |
CN114169234A (en) | Scheduling optimization method and system for unmanned aerial vehicle-assisted mobile edge calculation | |
CN111757361A (en) | Task unloading method based on unmanned aerial vehicle assistance in fog network | |
Zhao et al. | Online distributed optimization for energy-efficient computation offloading in air-ground integrated networks | |
CN113950081A (en) | Dynamic service migration and request routing method facing microservice in multi-unit mobile edge computing | |
CN114630397B (en) | Unmanned aerial vehicle access selection method based on time slot division | |
Tang et al. | Completed tasks number maximization in UAV-assisted mobile relay communication system | |
CN113627013A (en) | System throughput maximization method based on unmanned aerial vehicle binary unloading edge calculation | |
CN112579290A (en) | Unmanned aerial vehicle-based calculation task migration method for ground terminal equipment | |
Wang et al. | Joint optimization of resource allocation and computation offloading based on game coalition in C-V2X | |
Zeng et al. | Joint resource allocation and trajectory optimization in UAV–enabled wirelessly–powered MEC for large area | |
Wang et al. | Vehicular computation offloading in UAV-enabled MEC systems | |
CN116723548A (en) | Unmanned aerial vehicle auxiliary calculation unloading method based on deep reinforcement learning | |
Birabwa et al. | Multi-agent deep reinforcement learning for user association and resource allocation in integrated terrestrial and non-terrestrial networks | |
CN116634466A (en) | Unmanned aerial vehicle-based collaborative multi-access edge computing task unloading and resource allocation method | |
CN116366127A (en) | Task completion rate maximization method for unmanned aerial vehicle auxiliary multi-MEC server | |
Lee et al. | Multi-Agent Reinforcement Learning in Controlling Offloading Ratio and Trajectory for Multi-UAV Mobile Edge Computing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |