A kind of NOMA honeycomb car networking dynamic resource scheduling method based on efficiency
Technical field
The invention belongs to mobile communication fields, are related to a kind of NOMA honeycomb car networking dynamic resource scheduling side based on efficiency
Method.
Background technique
In recent years, intelligent transportation system (ITS) constantly receives significant attention, and honeycomb car networking communicates (C-V2X), as one
Kind car networking mainstream technology realizes vehicle and infrastructure-based communication (C-V2I) and vehicle and Che Zhi using existing cellular telecommunication art
Letter (C-V2V) is connected, traffic efficiency, the reliable low time delay demand of superelevation such as following automatic Pilot of more potential satisfaction are not only improved.
In the communication of dense city magnanimity vehicle, frequency spectrum resource is very short, on the one hand, limited frequency spectrum is made full use of to provide
Source, designing effective resource optimization algorithm can be improved the handling capacity of system, meets the demands such as different user QoS.On the other hand,
By introducing NOMA technology, more phone users can be linked into network simultaneously, have apparent performance excellent intensive scene
Gesture is more suitable for the system deployment in the following city, can promote spectrum efficiency and throughput of system.
In order to meet the mobile lower highly reliable and low time delay demand communicated of vehicle high-speed, 3GPP (3rd Generation
Partnership Project) the enhancing V2V communication technology towards vehicle communication is proposed on D2D communication infrastructure, not only alleviate
Load of base station overweight status, also reduces propagation delay time.Simultaneously in practice, data amount of reach is random and unknown, when
When data amount of reach is more than that network allows to access data volume, need to control accessible data volume to avoid network congestion.
Find that it has the following disadvantages: in the research process to the prior art
Firstly, early-stage study mainly considers the resource allocation problem of V2V communication under OFDMA system, limiting allows to connect
Enter phone user's number of network.Under the NOMA cellular network scene for supporting V2V communication, extensive work concentrates on maximizing frequency spectrum
Efficiency and handling capacity maximize the resource point of system energy efficiency while not considering to extend to reliability when guaranteeing V2V user
With problem.Also, in practice, data amount of reach be easy to cause network congestion more than the threshold value of network capacity, and most of
Congestion control is not integrated in resource optimization model by research.Finally, existing literature support V2V communication NOMA scene under,
Consider to establish static Optimized model mostly, cannot real-time dynamicly carry out scheduling of resource according to network load state.
Therefore, the present invention support V2V communication NOMA cellular network scene under, fully consider V2V user when extend to
The restrictive conditions such as reliability, the rate requirement of NOMA user, string stability, the control of access data volume and user power control, needle
To the power distribution problems in the co-channel interference and NOMA criterion of V2V user and NOMA user, construct to maximize system energy efficiency
For the optimization problem of target, a kind of dynamic resource allocation for combining subchannel scheduling, power control and congestion control is finally proposed
Algorithm.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of NOMA honeycomb car networking dynamic resource scheduling based on efficiency
Method maximizes system energy efficiency under the premise of guaranteeing system stability.
In order to achieve the above objectives, the invention provides the following technical scheme:
A kind of NOMA honeycomb car networking dynamic resource scheduling method based on efficiency, which is characterized in that this method are as follows: propping up
Under the NOMA cellular network scene for holding V2V communication, according to the reliability of V2V user, V2V user's time delay, NOMA user rate need
Ask and the power limit of user be constraint condition, using maximize system energy efficiency it is long when average efficiency as optimization aim, and be directed to
The power distribution problems under interference and NOMA criterion between V2V user and phone user establish the son of joint NOMA user
Channel distribution, V2V user frequency spectrum distribution and congestion control demand Stochastic Optimization Model, be NOMA user and V2V user's system
Determine power distribution and subchannel scheduling strategy.
Further, meet the reliability requirement of the V2V user are as follows: V2V user causes in shared NOMA user's subchannel
Interference will reduce V2V user communication quality, by using bit error rate (Bit Error Rate, BER) constrain guarantee V2V
The communication quality of user mitigates signal interruption and packet loss as caused by interference etc.;
Meet V2V user's delay requirement are as follows: V2V communication carrier delay sensitive business, usually transmit vehicle driving and
The relevant security information of road traffic, delay requirement constraint with to avoid user's transmission process since cause need not for the factors such as interference
The packet loss wanted or transmission delay;
Meet the rate requirement of the NOMA user are as follows: interference pair caused by order to control V2V in user sharing frequency spectrum
The influence of the communication link quality of NOMA user;
The power demand of NOMA user and V2V user are as follows: share the power of the NOMA user of same subchannel and be no more than
Its threshold value, with the V2V user power of the same subchannel of NOMA user sharing and also no more than its threshold value.
Further, the subchannel distribution demand of the NOMA user are as follows: phone user uses NOMA Techno-sharing subchannel
When, to guarantee communication quality, it must not exceed maximum multiplexing number M;
The frequency spectrum distribution requirements of the V2V user are as follows: although multiple users are multiplexed same phone user's frequency spectrum and can be improved
Spectrum efficiency, but NOMA user can be interfered when V2V user sharing frequency spectrum, usually assume that V2V user is at most one shared
NOMA user's subchannel;
The congestion control demand are as follows: when data packet amount of reach is more than that network allows to access data volume, connect by control
Enter the data volume of network and improve data transfer rate and guarantees string stability to avoid network congestion.
Further, the buffer queue renewal process of the NOMA user in the NOMA cellular network over each slot are as follows:
Qi(t+1)=max { Qi(t)-ri(t),0}+Γi(t)
Wherein, Qi(t+1) queue length of i-th of NOMA user when next time slot starts is indicated;Qi(t) i-th is indicated
Queue length of a NOMA user when current time slots start;Γi(t) indicate that i-th of NOMA user allows in current time slots
The data volume of access;ri(t) the data packet number that i-th of phone user leaves in current time slots is indicated.
Further, the string stability of each NOMA user are as follows:
Wherein,Indicate the time average length of i-th of NOMA user;T indicates that i-th of NOMA user is lined up the period;E table
Show in whole cycle T, the queue length of NOMA user i in system is averaged.
Further, Lyapunov function characterizes the queue congestion degree of system, and functional value is bigger, and queue is longer, user
The time that transmission data need to wait is also longer.The queue vector of t slotted system herein is expressed as Q (t)=[Qi(t),Qk
(t),Hi(t)], optimization problem is considered as maximizing rate and the time of minimum power is average, therefore can use Li Yapu
Nuo Fu is deviated with the upper bound of the sum of weighting cost function and is carried out the selection of control strategy, so that optimal power distribution is obtained,
Reached while guaranteeing string stability, the maximization network time averagely under user rate purpose.Define Lyapunov
Function are as follows:
Wherein, QiIt (t) is the NOMA customer service queue at current time, HiIt (t) is the virtual team of current time NOMA user
Column, QkIt (t) is V2V user's virtual queue at current time.
Further, the optimization aim be classified into three steps respectively obtain congestion control, subchannel scheduling and power distribution it is excellent
Neutralizing can be obtained since congestion control problem is linear problem with direct solution, then design a kind of suboptimum subchannel matching
Algorithm obtains subchannel scheduling scheme, specific subchannel matching algorithm step are as follows: is each NOMA in each time slot scheduling
User and V2V user dynamically distribute subchannel, to meet each constraint condition, specific steps are as follows:
In each time slot scheduling, the wireless of all NOMA users and V2V user's current time slots is estimated by channel model
Channel state information, in the channel model of the user, phone user uses Rayleigh channel rapid fading model, considers V2V user
Fast moving property, 3D geometric space channel model more can the actual channel model of accurate simulation, using the city WINNER II believe
Road model.And observe the queue cache information and virtual queue status information of each user;
User is matched with subchannel according to the channel preference matrix of user, to maximize system energy efficiency, due to same
Reusable maximum number of user is less than M on one sub-channels, therefore acquires globally optimal solution, but its complexity with exhaustive search algorithm
Opposite number of users is in exponential increase, therefore proposes a kind of suboptimum user's scheduling scheme for reducing complexity.Initialising subscriber first
Power, it is assumed that each user distributes identical mean powerAndUser's collectionWithIt is initial respectively
Change the user for recording unallocated subchannel, in scheduling scheme, when the number of users for being multiplexed same subchannel is no more than threshold value M
When, first according to the channel preference matrix of NOMA userMatch optimum channel gain and corresponding by channel distribution
To user, and by element zero setting in respective channels matrix and update user's collectionSince NOMA user and V2V are with per family
Reusable channel, therefore search again for its corresponding channel preference matrixWithBy comparing
The user with maximum channel gain is found, and distributes corresponding channel, until the users multiplexing number of each channel reaches M, is obtained
Collect U to feasible usern,possible, collect the smallest user of objective function by calculating selection, the user of unassigned subchannel
Then return to user's collectionWithAlgorithm is executed until all users are assigned to subchannel by circulation.
Last dump power optimization problem, since the problem is that non-convex optimization problem is difficult to direct solution, benefit in the present invention
Optimization problem is converted with convex row approximation theory, and obtains power distribution strategies by Lagrangian decomposition scheme, is had
Body power allocation scheme solution procedure are as follows:
Vector approximation value is initialized, Lagrange multiplier is initialized, the queue length and V2V of the initial NOMA user is used
The virtual queue length at family carries out successive ignition until meeting according to convex row approximate algorithm according to the user resources amount of initialization
Until the condition of convergence, the optimization solution and the vector approximation value after optimization for obtaining power, former non-convex optimization problem is converted into convex excellent
Change problem;Using obtained power as the initialization power of algorithm 3, successive ignition updates Lagrange multiplier and convex row is excellent
Power policy after changing conversion.By iteration for several times, judge whether to meet the condition of convergence, if the relaxation of front and back iteration twice
The absolute value of the difference of target function value afterwards is less than or equal to the given limits of error, or has reached maximum number of iterations, then
It terminates iterative process and takes the resulting power distribution result of last time iteration as the optimal power allocation strategy of current time slots,
If convex row approximate algorithm and power distribution algorithm are all satisfied the condition of convergence, the power distribution strategies are distributed to all
User;According to resource allocation policy, all users update queue length according to buffer queue and virtual queue more new formula, wait
Next time slot scheduling.
The beneficial effects of the present invention are:
The present invention is under the NOMA cellular network scene for supporting V2V communication, for dry between V2V user and phone user
Disturb and NOMA criterion under power distribution problems, establish joint subchannel dispatch, power distribution and congestion control it is random excellent
Change model, maximizes system energy efficiency while guaranteeing V2V user's time delay, reliability and phone user's rate requirement.It utilizes
Liapunov randomized optimization process carries out dynamic resource scheduling according to the load condition of current network, by Stochastic Optimization Model
Three subproblems that single time slot solves are converted and be decomposed into, can access data volume guarantee string stability by controlling to avoid net
Network congestion, further designs a kind of suboptimization user and subchannel matching algorithm obtains subchannel scheduling scheme.Finally, using connecting
Continue convex approximation theory and convert convex optimization problem for power distribution subproblem, and power point is obtained using Lagrange duality method
With strategy.
Detailed description of the invention
In order to keep the purpose of the present invention, technical scheme and beneficial effects clearer, the present invention provides following attached drawing and carries out
Illustrate:
Fig. 1 is the vehicle communication scene figure under dense city honeycomb car networking;
Fig. 2 is suboptimum subchannel dispatching algorithm flow chart;
Fig. 3 is that the iterative power allocation algorithm flow chart decomposed with Lagrange duality is approached based on convex row;
Fig. 4 is that the NOMA federated resource based on efficiency dispatches Global Algorithm flow chart.
Specific embodiment
Below in conjunction with attached drawing, a preferred embodiment of the present invention will be described in detail.
Shown in Figure 1, Fig. 1 is the vehicle communication scene figure under the dense city honeycomb car networking that the present invention uses.?
In present example, the transmitting scene including two kinds of communication modes of V2I downlink and V2V skidding is considered.In downlink V2I communication, lead to
Cross and introduce the number of devices that NOMA technology increases access network, V2V skidding direct communication then alleviates the burden of base station, and have compared with
Low end-to-end propagation delay time.Downtown roads are latticed in Manhattan, and one-dimensional Poisson points distribution models are obeyed in vehicle distributionWhat the wireless channel model of the user was made of path loss, shadow fading declines slowly
It falls and is constituted with rapid fading.NOMA user uses Rayleigh channel rapid fading model, considers the fast moving property of V2V user, 3D geometry
Spatial Channel Model more can the actual channel model of accurate simulation, referring to 3GPP use the city WINNER II fast fading channel mould
Type.
Shown in Figure 2, suboptimum subchannel dispatching algorithm flow chart, target is obtained by the subchannel matching algorithm
The channel dispatch strategy of user.Steps are as follows:
Step 201: initialising subscriber power observes customer service quene state and virtual queue state.
Step 202: constructing the channel gain matrix Η of NOMA user and V2V user respectivelyiAnd Ηk。
Step 203: from channel matrix ΗiMiddle search maximum channel gain simultaneously carries out corresponding channel dispatch.
Step 204: judging whether to meet loop termination condition, i.e.,If satisfied, then terminating calculation
Method exports subscriber channel scheduling scheme.It is unsatisfactory for, continues step 205.
Step 205: further respectively from channel matrix ΗiAnd ΗkMiddle search maximum channel gain simultaneously compares to obtain maximum
Person carries out corresponding channel dispatch.
Step 206: the scheduling solution of user being calculated according to formula and updates user and dispatches collectionWith
Step 207: judging whether to meet loop termination condition, i.e.,If satisfied, then terminating calculation
Method exports subscriber channel scheduling scheme.It is unsatisfactory for, continues step 205.
Shown in Figure 3, Fig. 3 is that the iterative power allocation algorithm stream decomposed with Lagrange duality is approached based on convex row
Cheng Tu, steps are as follows:
Step 301: initialization the number of iterations N1With limits of error Δ1, initialize feasible point, i.e. NOMA user and V2V
User power, initialization the number of iterations index.
Step 302: the power of initialization being substituted into calculating formula and obtains vector approximation.
Step 303: judging whether to meet loop convergence condition, if not satisfied, then termination algorithm, output introduces convex optimization reason
Optimal power solution after.Continue step 304 if meeting.
Step 304: by the power of update substitute into respectively calculating formula acquire update after vector approximation.
Step 305: judging whether to meet loop convergence condition, if not satisfied, then termination algorithm, output introduces convex optimization reason
Optimal power solution after.Continue step 304 if meeting.
Step 306: vector approximation substitution optimization problem will be updated and solved, update current optimal power solution, and gradually
Increase the number of iterations.
Step 307: initialization approximation vector, Lagrange multiplier ν0,λ0,u0,η0, maximum number of iterations N2, convergence item
Part and iteration index etc..
Step 308: in current time slots, service queue state, virtual queue and the estimation for observing the user of the time slot should
The channel state information of time slot.
Step 309: judging whether to meet loop convergence condition, if not satisfied, terminating cycling condition, output power distribution side
Case.
Step 310: if satisfied, a preceding iterative power and Lagrange multiplier are substituted by derivation formula by KKT condition,
Acquire the power distribution optimal policy of current iteration.
Step 311: Lagrange multiplier ν is updated according to Subgradient Algorithmm,λm,um,ηmAnd the number of iterations.
Step 312: judging whether to meet loop convergence condition.If not satisfied, cycling condition is terminated, the function after output optimization
Rate allocation plan.
It is shown in Figure 4, Global Algorithm flow chart is dispatched for the NOMA federated resource based on efficiency, steps are as follows:
Step 401: slot length is arranged in initialization control parameter V, NOMA Subscriber Queue length and virtual queue length.
Step 402: current time slots are judged whether in setting slot range, if thening follow the steps in slot range
403;Otherwise algorithm terminates.
Step 403: observing the NOMA quene state, virtual queue length and the channel status for estimating the time slot of the time slot
The linear solution of congestion control is obtained by calculation in information.
Step 404: executing suboptimization subchannel distribution algorithm and obtain the channel dispatch strategy of NOMA user and V2V user.
Step 405: execute based on convex row approach with Lagrange duality decompose iterative power allocation algorithm obtain it is excellent
The power allocation scheme of change.
Step 406: next time slot NOMA Subscriber Queue state and virtual queue length are updated according to queue more new formula.
Step 407: going to next time slot and continue above step.
Finally, it is stated that preferred embodiment above is only used to illustrate the technical scheme of the present invention and not to limit it, although logical
It crosses above preferred embodiment the present invention is described in detail, however, those skilled in the art should understand that, can be
Various changes are made to it in form and in details, without departing from claims of the present invention limited range.