CN104168661A

CN104168661A - Transmission scheduling method for network lifetime maximization that satisfies fairness condition

Info

Publication number: CN104168661A
Application number: CN201410275674.2A
Authority: CN
Inventors: 胡封晔; 尹颖奇; 杜大鲲; 臧达霏; 邓云蕾; 赵利英; 王志军; 杜宇; 王丰; 王璐
Original assignee: Jilin University
Current assignee: Jilin University
Priority date: 2014-06-19
Filing date: 2014-06-19
Publication date: 2014-11-26
Anticipated expiration: 2034-06-19
Also published as: CN104168661B

Abstract

本发明涉及一种满足公平性条件的最大化网络生存期传输调度方法，属于无线通信领域。包括将传感器节点选择的问题建模为约束马尔可夫决策过程，利用动态规划中的Bellman方程求解CMDP问题，减少计算复杂度得出最终的约束最优生存期和最优策略，确定每个时隙选择的传感器节点。本发明针对无线体域网在实际应用中需要满足一定的公平性的情况，克服了传统传输调度方法仅考虑最大化生存期而忽略了公平性的缺点，在保证满足公平性约束条件下最大化了网络的生存期，更加适用于对一个病人监测多项生理参数的情况。The invention relates to a transmission scheduling method for maximizing the network lifetime satisfying fairness conditions, and belongs to the field of wireless communication. Including modeling the problem of sensor node selection as a constrained Markov decision process, using the Bellman equation in dynamic programming to solve the CMDP problem, reducing the computational complexity to obtain the final constrained optimal lifetime and optimal strategy, and determining each time Gap selected sensor nodes. The invention aims at the situation that the wireless body area network needs to meet certain fairness in practical application, overcomes the shortcomings of the traditional transmission scheduling method that only considers maximizing the lifetime and ignores the fairness, and maximizes The lifetime of the network is improved, and it is more suitable for monitoring multiple physiological parameters of a patient.

Description

A Transmission Scheduling Method for Maximum Network Lifetime Satisfied with Fairness Conditions

技术领域 technical field

本发明涉及无线通信领域，是一种面向无线体域网的信号传输调度方法。 The invention relates to the field of wireless communication, and relates to a signal transmission scheduling method oriented to a wireless body area network. the

背景技术 Background technique

无线体域网(WBAN,Wireless Body Area Network)作为物联网的重要组成部分和无线传感器网络在医疗监护等方面的应用，逐渐成为了人们关注的焦点。因为组成无线体域网的传感器节点由电池供电，能量受到严格的制约，所以设计高效、公平的传输调度方法来延长网络生存期是体域网的研究重点和热点，不仅具有理论研究意义，还具有重要的实际应用价值。 Wireless Body Area Network (WBAN, Wireless Body Area Network), as an important part of the Internet of Things and the application of wireless sensor networks in medical monitoring, has gradually become the focus of attention. Because the sensor nodes that make up the wireless body area network are powered by batteries, the energy is strictly restricted, so designing an efficient and fair transmission scheduling method to prolong the network lifetime is the focus and hotspot of body area network research, which not only has theoretical research significance, but also It has important practical application value. the

Dagher等人提出了一种最大化网络生存期的迭代算法，把最大化网络生存期的问题归结为一个PO(Pareto-Optimal)问题并且得出了最优解。Chen等提出了无线传感器网络中生存期的一般化公式，并且证明了信道状况信息和剩余能量信息是影响网络生存期的主要因素。随后Chen等又提出一个被称为DPLM(Dynamic Protocol for Lifetime Maximization)的动态传输调度协议并且取得了渐进最优的网络生存期。Cohen等提出了一种时变机会协议，仿真结果表明该协议获得的生存期优于与其对比的其它几种协议。Madan等提出了一种结合了物理层、MAC层和路由层的联合优化算法，该算法将计算最优路径流、链路调度和传输功率归纳成一个非线性最优化问题，大大延长了网络生存期。Zhai等运用协作通信的方法设计了一种新的协作MAC协议，有效降低了发射功率，提高了无线传感器网络的生存期。2012年，Movassaghi等提出一个热量和能量自感应的体域网能效路由协议，根据节点的温度和能量水平为每个路径计算成本函数来决定路由的选择，有效地利用可用资源，延长了体域网的生存期。Ortiz等提出了一种基于多跳树的自适应路由协议，构建了一个同时考虑节点电池余量、接收信号强度和跳数的生成树，协议能够平衡节点间的能量消耗，延长网络使用寿命。刘汉春等以延长体域网网络生存期为目标提出了一种基于加入临时节点的高能效路由算法。该算法针对体域网在实际应用中出现的新情况，通过充分利用临时节点的剩余能量平衡网络中各传感器节点之间能量消耗，仿真结果表明，该算法大大延长了网络的生存期。 Dagher et al. proposed an iterative algorithm for maximizing the network lifetime, reduced the problem of maximizing the network lifetime to a PO (Pareto-Optimal) problem and obtained the optimal solution. Chen et al. proposed a generalized formula for the lifetime of wireless sensor networks, and proved that channel status information and remaining energy information are the main factors affecting the lifetime of the network. Then Chen et al. proposed a dynamic transmission scheduling protocol called DPLM (Dynamic Protocol for Lifetime Maximization) and achieved an asymptotically optimal network lifetime. Cohen et al. proposed a time-varying chance protocol, and the simulation results show that the lifetime obtained by this protocol is better than other protocols compared with it. Madan et al. proposed a joint optimization algorithm that combines the physical layer, MAC layer, and routing layer. This algorithm summarizes the calculation of optimal path flow, link scheduling, and transmission power into a nonlinear optimization problem, which greatly prolongs the network survival. Expect. Zhai et al. designed a new cooperative MAC protocol by using the method of cooperative communication, which effectively reduces the transmission power and improves the lifetime of wireless sensor networks. In 2012, Movassaghi et al. proposed a heat and energy self-sensing body area network energy-efficient routing protocol, which calculates the cost function for each path according to the temperature and energy level of the node to determine the routing selection, effectively utilizes available resources, and extends the body area network. The lifetime of the network. Ortiz et al. proposed an adaptive routing protocol based on multi-hop trees, and constructed a spanning tree that considered node battery remaining, received signal strength and hop count at the same time. The protocol can balance the energy consumption between nodes and prolong the service life of the network. Liu Hanchun et al. proposed an energy-efficient routing algorithm based on adding temporary nodes with the goal of prolonging the lifetime of the body area network. Aiming at the new situation in the practical application of body area network, the algorithm balances the energy consumption between sensor nodes in the network by making full use of the remaining energy of temporary nodes. The simulation results show that the algorithm greatly prolongs the lifetime of the network. the

网络的资源应公平的分配给各传感器节点，公平性是无线体域网调度方法中的关键问题之一。公平的调度方法应根据各传感器节点的特性需求分配相应的资源，防止某个节点占用过多的信道资源而影响到其他用户的传输。无线体域网对实时性的要求很高，如果某个传感器节点占用过多的信道资源而导致其他节点长时间得不到机会传输，对重要的生理参数预警不及时会导致很严重的后果。好的传输调度方法不仅要最大化整个网络的生存期，同时也应保证每个传感器节点的公平性。上述各种最大化网络生存期的方法均没有考虑到公平性的要求，不能完全满足实际应用中多种传感器节点对资源需求不同的情况。经过对现有技术文献的检索，尚未见到将生存期与公平性联合考虑的方法的相关报道。 The resources of the network should be allocated to each sensor node fairly, and fairness is one of the key issues in the scheduling method of wireless body area network. A fair scheduling method should allocate corresponding resources according to the characteristic requirements of each sensor node, so as to prevent a node from occupying too many channel resources and affecting the transmission of other users. Wireless body area networks have high requirements for real-time performance. If a sensor node occupies too many channel resources, other nodes will not get the opportunity to transmit for a long time, and the early warning of important physiological parameters will lead to serious consequences. A good transmission scheduling method should not only maximize the lifetime of the entire network, but also ensure the fairness of each sensor node. All the above-mentioned methods of maximizing the network lifetime do not take into account the requirement of fairness, and cannot fully satisfy the situation that various sensor nodes have different resource requirements in practical applications. After searching the literature of the prior art, there is no relevant report on the method that considers the survival period and the fairness jointly. the

发明内容 Contents of the invention

本发明提供一种满足公平性条件的最大化网络生存期传输调度方法，以解决目前各种最大化网络生存期的方法均没有考虑到公平性的要求，不能完全满足实际应用中多种传感器节点对资源需求不同的情况的问题。 The present invention provides a transmission scheduling method for maximizing the network lifetime that satisfies fairness conditions, so as to solve the problem that the current methods for maximizing the network lifetime do not take into account the requirement of fairness, and cannot fully meet the requirements of various sensor nodes in practical applications. Problems with situations with different resource requirements. the

本发明采取的技术方案是，包括 The technical scheme that the present invention takes is, comprises

(1)无线体域网，该无线体域网是一个以人体为监测对象的网络，将若干个功能不同的传感器节点置于体表或者体内的相应位置，周期地监测和记录各种生理信息，各传感器节点将采集的数据通过蓝牙、Zigbee、超宽带或其他方式的人体通信技术传递给汇聚节点AP(Access Point)，AP再将这些信息通过外部网络传送到远程控制中心； (1) Wireless body area network, which is a network that takes the human body as the monitoring object. Several sensor nodes with different functions are placed on the body surface or corresponding positions in the body to periodically monitor and record various physiological information. , each sensor node transmits the collected data to the aggregation node AP (Access Point) through Bluetooth, Zigbee, ultra-wideband or other human body communication technologies, and the AP transmits the information to the remote control center through the external network;

在IEEE802.15.6通信标准的MAC层协议中，时间被分成了等长的超帧结构，超帧结构包括如下四个部分：控制阶段、竞争接入阶段CAP、竞争空闲阶段CFP、非活跃阶段，CFP阶段又继续分成若干时隙，在无线体域网中，时间是待分配给各个传感器节点的资源，无线体域网中所有传感器节点共用一个信道，在每个时隙只能有一个传感器节点将采集的数据传送到AP节点，被选中的传感器节点消耗传输数据所需的能量； In the MAC layer protocol of the IEEE802.15.6 communication standard, time is divided into a superframe structure of equal length. The superframe structure includes the following four parts: control phase, contention access phase CAP, contention idle phase CFP, inactive phase, The CFP stage continues to be divided into several time slots. In the wireless body area network, time is the resource to be allocated to each sensor node. All sensor nodes in the wireless body area network share a channel, and there can only be one sensor node in each time slot. Transmit the collected data to the AP node, and the selected sensor node consumes the energy required to transmit the data;

(2)将传感器节点选择的问题建模为约束马尔可夫决策过程，一个传输调度协议恰好对应CMDP问题中的一个策略u，将从状态i开始最大的网络生存期为L^*(i)，表示如下： (2) The problem of sensor node selection is modeled as a constrained Markov decision process. A transmission scheduling protocol corresponds to a policy u in the CMDP problem. The maximum network lifetime from state i is L ^* (i), Expressed as follows:

${L L}^{* *} ((i i)) = = \underset{u u}{max max} {L L}_{u u} ((i i))$

如果网络使用某个策略u在达到停止状态之前获得的报酬最多(即网络生存期最长)，那么我们称这个策略为最优策略，用u^*表示， If the network uses a certain strategy u to get the most rewards before reaching the stop state (that is, the network has the longest lifetime), then we call this strategy the optimal strategy, denoted by u ^* ,

${L L}_{{u u}^{* *}} ((i i)) = = {L L}^{* *} ((i i)),, &ForAll; &ForAll; i i &Element; &Element; S S \ \ {S S}_{t t}$

定义U_a＝{u∈U:f≥f_n}为可行策略集。如果策略u^*∈U_a满足条件L(u^*)≥L(u),(u∈U_a)，则u^*被称为约束最优策略； Define U _a ={u∈U:f≥f _n } as a set of feasible strategies. If the strategy u ^* ∈U _a satisfies the condition L(u ^* )≥L(u),(u∈U _a ), then u ^* is called the constrained optimal strategy;

CMDP的目标是寻找一个最优的策略u^*∈U_a来最大化网络生存期L，最优的传输调度协议由最优策略u^*给出，即明确了最优策略u^*就可以知道在每个时隙选择哪个传感器节点来发送数据； The goal of CMDP is to find an optimal policy u ^* ∈ U _a to maximize the network lifetime L. The optimal transmission scheduling protocol is given by the optimal policy u ^* , that is, the optimal policy u ^* can be known in Which sensor node is selected for each time slot to send data;

CMDP模型中的各个参数介绍如下： The parameters in the CMDP model are introduced as follows:

(a)状态空间 (a) State space

在每个时隙，网络的状态i由剩余能量e，传输所需要的能量w和公平系数f共同组成。我们定义网络状态空间S如下： In each time slot, the state i of the network is composed of the remaining energy e, the energy w required for transmission and the fairness coefficient f. We define the network state space S as follows:

S＝{i＝(e,w,f)} S={i=(e,w,f)}

当网络生存期耗尽时，网络达到一个特殊的终止状态S_t如下： When the network lifetime is exhausted, the network reaches a special termination state S _t as follows:

${S S}_{t t} = = {{((e e,, w w,, f f)) : : &ForAll; &ForAll; {{{e e}_{n no} : : {e e}_{n no} < < {ϵ ϵ}_{11}}} ore ore < < w w}}$

e_n为传感器节点n的剩余能量，ε₁为一个时隙传感器节点采用最小发射功率传输数据需要的能量，e_n:e_n<ε₁表示传感器节点n的剩余能量在任何的信道状况下都不能完成一次传输，e<w意味着一次传输失败； e _n is the remaining energy of sensor node n, ε ₁ is the energy required for a time slot sensor node to transmit data with the minimum transmission power, e _n : e _n <ε ₁ means that the remaining energy of sensor node n is under any channel condition A transfer cannot be completed, e<w means a transfer failed;

(b)行动空间 (b) Action space

用A来表示行动的集合，当网络处于状态i＝(e,w,f)∈S时，行动空间可以表示为： Use A to represent the set of actions. When the network is in the state i=(e,w,f)∈S, the action space can be expressed as:

A(i)＝A[(e,w,f)]＝{n:e_n≥w_n} A(i)=A[(e,w,f)]={n:e _n ≥w _n }

根据终止状态的定义，可以得出，任何非终止状态的行动空间均为非空的； According to the definition of terminal state, it can be concluded that the action space of any non-terminal state is non-empty;

(c)转移概率 (c) Transition probability

当网络的状态为i，经过行动a的作用后，下一个状态为j的概率为 When the state of the network is i, after action a, the probability of the next state being j is

${P P}_{iaj iaj} = = p p (({w w}^{' '})) 11_{[[{e e}^{' '} = = e e - - {I I}_{n no} {w w}_{n no}]]}$

其中p(w')＝Pr{W＝w'}是W的概率密度函数，是由信道衰落决定的； Where p(w')=Pr{W=w'} is the probability density function of W, which is determined by channel fading;

(d)传输报酬 (d) Transfer Remuneration

在每次数据传输之后，网络都会得到一个单元的回报，直到网络进入停止状态，也就是说，如果网络的状态为这个时隙的立即回报即为 After each data transfer, the network is rewarded with one unit until the network enters a stalled state, that is, if the state of the network is The immediate reward for this time slot is

$R R ((i i)) = = 11_{[[i i &Element; &Element; S S \ \ {S S}_{T T}]]}$

(e)约束条件 (e) Constraints

假设网络处于状态i，如果经过行动a的作用以后网络的公平系数f大于等于给定的阈值f_n，则我们称行动a为可行的行动，公式可表示如下： Assuming that the network is in state i, if the fairness coefficient f of the network after action a is greater than or equal to a given threshold f _n , then we call action a a feasible action, and the formula can be expressed as follows:

f(i,a)≥f_n f(i,a)≥f _n

f(i,a)表示网络在状态i经过a的作用后到达下一个状态时的公平系数，网络在状态i可行的行动集由A_a(i)来表示； f(i,a) represents the fairness coefficient of the network when it reaches the next state after the action of a in state i, and the feasible action set of the network in state i is represented by A _a (i);

(f)策略 (f) strategy

策略集U中的策略u是一个序列u＝{u₀,u₁,...}，其中u_n:S→{1,...,N}表示在第n个时隙选择的传感器节点，u_n(i₁,a₁,i₂,a₂,...,i_n-1,a_n-1,i_n)是A的条件概率测度，以首个节点能量耗尽的时刻度量生存期，所以，整个网络的生存期L可以由网络到达停止状态St之前所有的回报之和来描述，定义L_u(i)为网络从状态i开始，使用策略u的生存期，即所有回报之和； The policy u in the policy set U is a sequence u={u ₀ ,u ₁ ,...}, where u _n :S→{1,...,N} represents the sensor node selected in the nth time slot , u _n (i ₁ ,a ₁ ,i ₂ ,a ₂ ,...,i _n-1 ,a _n-1 ,i _n ) is the conditional probability measure of A, measured at the moment when the energy of the first node is exhausted Therefore, the lifetime L of the entire network can be described by the sum of all rewards before the network reaches the stop state St. Define L _u (i) as the lifetime of the network starting from state i and using strategy u, that is, all rewards Sum;

(3)求解最优策略 (3) Solve the optimal strategy

从状态i开始，满足约束条件的最大的网络生存期L^*(i)是如下Bellman最优方程的唯一解， Starting from state i, the maximum network lifetime L ^* (i) that satisfies the constraints is the only solution to the following Bellman optimal equation,

${L L}^{* *} ((i i)) = = {L L}^{* *} [[((e e,, w w,, f f))]] = = R R ((i i)) \underset{n no &Element; &Element; A A ((i i))}{max max} {{\underset{j j &Element; &Element; S S}{Σ Σ} {p p}_{iaj iaj} {L L}^{* *} ((j j))}},, &ForAll; &ForAll; i i &Element; &Element; S S$

s.t.f_j≥f_n stf _j ≥ f _n

其中f_j是网络在状态j下的公平系数，f_n是根据不同的应用场景给出的公平阈值； Where f _j is the fairness coefficient of the network in state j, and f _n is the fairness threshold given according to different application scenarios;

实际上，上面的公式还可以写作 In fact, the above formula can also be written as

${L L}^{* *} ((i i)) = = {L L}^{* *} [[((e e,, w w,, f f))]] = = R R ((i i)) \underset{n no &Element; &Element; {A A}_{a a} ((i i))}{max max} {{\underset{j j &Element; &Element; S S}{Σ Σ} {p p}_{iaj iaj} {L L}^{* *} ((j j))}},, &ForAll; &ForAll; i i &Element; &Element; S S$

传输调度方案的最优策略u可以由如下公式得出 The optimal policy u of the transmission scheduling scheme can be obtained by the following formula

$u u ((i i)) = = \underset{n no &Element; &Element; {A A}_{a a} ((i i))}{arg arg max max} {{{\underset{j j &Element; &Element; S S}{Σ Σ} {p p}_{iaj iaj} L L}^{* *} ((j j))}},, &ForAll; &ForAll; i i &Element; &Element; S S \ \ {S S}_{t t}$

(4)减少计算复杂度 (4) Reduce computational complexity

一个等效的Bellman最优方程写成如下形式： An equivalent Bellman optimality equation is written as follows:

$L L [[e e,, f f]] = = \underset{w w}{Σ Σ} p p ((w w)) {{R R [[e e,, w w]] + + \underset{n no &Element; &Element; {A A}_{a a}}{max max} L L [[e e - - {I I}_{n no} {w w}_{n no}]]}}$

所以，约束最优策略也可以通过下式求出 Therefore, the constrained optimal strategy can also be obtained by the following formula

$u u [[((e e,, w w,, f f))]] = = arg arg \underset{n no &Element; &Element; {A A}_{a a}}{max max} L L [[e e - - {I I}_{n no} {w w}_{n no}]]$

在实现过程中，每个时隙开始的时候，AP节点广播一个信标信号来唤醒所有传感器节点，为了使AP节点获取全局信道信息，所有的传感器节点都需要发送一个导频信号来回复AP的信标信号，AP节点利用接收到的信号来估计所有节点的信道状况并且得到传输需要的能量w，然后根据状态(e,w,f)来计算最优策略u，明确了最优策略，即知晓了在每个时隙应该选择的传感器节点，最后，AP节点广播被选中传感器节点的ID号，该传感器节点开始传输其采集到的数据，由于AP节点知道所有传感器节点的信道状况，所以能够继续更新网络状态为下一个时隙的到来做准备。 In the implementation process, at the beginning of each time slot, the AP node broadcasts a beacon signal to wake up all sensor nodes. In order for the AP node to obtain global channel information, all sensor nodes need to send a pilot signal to reply to the AP’s The beacon signal, the AP node uses the received signal to estimate the channel conditions of all nodes and obtain the energy w required for transmission, and then calculates the optimal strategy u according to the state (e, w, f), and defines the optimal strategy, namely Knowing the sensor node that should be selected in each time slot, finally, the AP node broadcasts the ID number of the selected sensor node, and the sensor node starts to transmit the data it collects. Since the AP node knows the channel status of all sensor nodes, it can Continue to update the network status to prepare for the arrival of the next time slot. the

本发明所述公平系数f定义如下：定义(T₁,T₂,...,T_N)表示每个传感器节点实际的传输次数，O表示已经传输的总次数，(b₁,b₂,...,b_N)是表示各个传感器节点重要性的权值。对于传感器节点n，定义标准化的传输次数为 The fairness coefficient f of the present invention is defined as follows: define (T ₁ , T ₂ ,..., T _N ) to represent the actual number of transmissions of each sensor node, O to represent the total number of transmissions already made, (b ₁ , b ₂ , ..., b _N ) are the weights representing the importance of each sensor node. For sensor node n, the normalized number of transmissions is defined as

x_n＝T_n/b_nO x _n =T _n /b _n O

假设有N个用户竞争网络资源，第n个用户获得的资源为x_n，那么定义网络的公平系数为 Assuming that there are N users competing for network resources, and the resource obtained by the nth user is x _n , then the fairness coefficient of the defined network is

$f f = = \frac{{((Σ Σ {x x}_{n no}))}^{22}}{NΣ NΣ {x x}_{n no}^{22}}$

f在0和1之间取值。如果网络的公平系数f＝1，则表示网络是完全公平的。网络的公平系数越大，公平程度越高，反之亦然。 f takes values between 0 and 1. If the fairness coefficient f=1 of the network, it means that the network is completely fair. The larger the fairness coefficient of the network, the higher the fairness, and vice versa. the

本发明将传感器节点选择的问题进行建模并归结为一个约束马尔可夫决策过程(CMDP,Constrained Markov Decision Processes)问题，提出了一种在公平性约束条件下最大化网络生存期的传输调度方法并且求解出公平性约束下的最优生存期和最优策略。 In the present invention, the problem of sensor node selection is modeled and attributed to a constrained Markov decision process (CMDP, Constrained Markov Decision Processes) problem, and a transmission scheduling that maximizes the network lifetime under the condition of fairness constraints is proposed method and solve the optimal lifetime and optimal strategy under the fairness constraints. the

本发明针对无线体域网在实际应用中需要满足一定的公平性的情况，克服了传统传输调度方法仅考虑最大化生存期而忽略了公平性的缺点，在保证满足公平性约束条件下采用最优策略最大化了网络的生存期，更加适用于对一个病人监测多项生理参数的情况。 Aiming at the situation that the wireless body area network needs to meet certain fairness in practical application, the present invention overcomes the shortcoming that the traditional transmission scheduling method only considers maximizing the lifetime while ignoring fairness, and adopts the best The optimal strategy maximizes the lifetime of the network and is more suitable for monitoring multiple physiological parameters of a patient. the

附图说明 Description of drawings

图1是本发明WBAN网络仿真模型； Fig. 1 is the WBAN network simulation model of the present invention;

图2是本发明无公平性要求下最优策略与其它协议生存期对比图； Fig. 2 is a comparison diagram of the optimal strategy and other protocol lifetimes without fairness requirements in the present invention;

图3是本发明传感器权重值相同时网络生存期随公平系数变化图； Fig. 3 is when the sensor weight value of the present invention is the same, the network survival period changes with the fair coefficient figure;

图4是本发明传感器权重值不同时网络生存期随公平系数变化图。 Fig. 4 is a graph showing the variation of the network lifetime with the fairness coefficient when the sensor weight values of the present invention are different. the

具体实施方式 Detailed ways

包括： include:

(1)无线体域网，该无线体域网是一个以人体为监测对象的网络，将若干个功能不同的传感器节点置于体表或者体内的相应位置，周期地监测和记录各种生理信息，各传感器节点将采集的数据通过蓝牙、Zigbee、超宽带或其他方式的人体通信技术传递给汇聚节点AP(Access Point)，AP再将这些信息通过外部网络传送到远程控制中心； (1) Wireless body area network, which is a network that takes the human body as the monitoring object. Several sensor nodes with different functions are placed on the body surface or corresponding positions in the body to periodically monitor and record various physiological information. , each sensor node transmits the collected data to the aggregation node AP (Access Point) through Bluetooth, Zigbee, ultra-wideband or other human body communication technologies, and the AP then transmits the information to the remote control center through the external network;

远程医疗监控是无线体域网最典型的应用，特别是用来连续监测和记录一些慢性疾病(如心脏病、糖尿病和哮喘等)患者的生理参数，通过安置在人体的传感器节点实时采集人体的各种生理参数，如血氧、脉搏、体温、心电等并传输到远程医疗中心来为监测的对象提供实时的健康指导。这样病人不会受到仪器线缆的束缚，扩大了活动空间。 Telemedicine monitoring is the most typical application of wireless body area network, especially for continuous monitoring and recording of physiological parameters of patients with chronic diseases (such as heart disease, diabetes and asthma, etc.), and real-time collection of human body parameters through sensor nodes placed on the human body. Various physiological parameters, such as blood oxygen, pulse, body temperature, ECG, etc., are transmitted to the telemedicine center to provide real-time health guidance for the monitored objects. In this way, the patient will not be bound by the instrument cables, which expands the room for movement. the

在IEEE802.15.6通信标准的MAC层协议中，时间被分成了等长的超帧结构，超帧结构包括如下四个部分：控制阶段、竞争接入阶段(CAP)、竞争空闲阶段(CFP)、非活跃阶段，CFP阶段又继续分成若干时隙，我们关注于数据包主要在CFP阶段传输的基于TDMA的协议，所以在无线体域网中，时间是待分配给各个传感器节点的资源，因为无线体域网中所有传感器节点共用一个信道，在每个时隙只能有一个传感器节点将采集的数据传送到AP节点，被选中的传感器节点消耗传输数据所需的能量，在每个时隙选择哪个传感器节点传输能够在满足节点公平性要求下最大化网络的生存期是研究中的关键问题； In the MAC layer protocol of the IEEE802.15.6 communication standard, the time is divided into a superframe structure of equal length. The superframe structure includes the following four parts: control phase, contention access phase (CAP), contention idle phase (CFP), In the inactive phase, the CFP phase continues to be divided into several time slots. We focus on the TDMA-based protocol where the data packets are mainly transmitted in the CFP phase. Therefore, in the wireless body area network, time is the resource to be allocated to each sensor node, because the wireless All sensor nodes in the body area network share one channel, and only one sensor node can transmit the collected data to the AP node in each time slot, and the selected sensor node consumes the energy required for data transmission. Which sensor node transmission can maximize the lifetime of the network under the requirement of node fairness is a key issue in the research;

${L L}^{* *} ((i i)) = = \underset{u u}{max max} {L L}_{u u} ((i i))$

(a)状态空间 (a) State space

S＝{i＝(e,w,f)} S={i=(e,w,f)}

(b)行动空间 (b) Action space

用A来表示行动的集合，当网络处于状态i＝(e,w,f)∈S时，行动空间可以表示为： Use A to represent the set of actions, when the network is in the state i=(e,w,f)∈S, the action space can be expressed as:

A(i)＝A[(e,w,f)]＝{n:e_n≥w_n} A(i)=A[(e,w,f)]={n:e _n ≥w _n }

根据终止状态的定义，可以得出，任何非终止状态的行动空间均为非空的； According to the definition of terminal state, it can be drawn that the action space of any non-terminal state is non-empty;

(c)转移概率 (c) Transition probability

(d)传输报酬 (d) Transfer Remuneration

$R R ((i i)) = = 11_{[[i i &Element; &Element; S S \ \ {S S}_{T T}]]}$

(e)约束条件 (e) Constraints

f(i,a)≥f_n f(i,a)≥f _n

(f)策略 (f) strategy

策略集U中的策略u是一个序列u＝{u₀,u₁,...}，其中u_n:S→{1,...,N}表示在第n个时隙选择的传感器节点，u_n(i₁,a₁,i₂,a₂,...,i_n-1,a_n-1,i_n)是A的条件概率测度，以首个节点能量耗尽的时刻度量生存期，所以，整个网络的生存期L可以由网络到达停止状态S_t之前所有的回报之和来描述，定义L_u(i)为网络从状态i开始，使用策略u的生存期，即所有回报之和； The policy u in the policy set U is a sequence u={u ₀ ,u ₁ ,...}, where u _n :S→{1,...,N} represents the sensor node selected in the nth time slot , u _n (i ₁ ,a ₁ ,i ₂ ,a ₂ ,...,i _n-1 ,a _n-1 ,i _n ) is the conditional probability measure of A, measured at the moment when the energy of the first node is exhausted Therefore, the lifetime L of the entire network can be described by the sum of all rewards before the network reaches the stop state S _t . Define L _u (i) as the lifetime of the network starting from state i and using strategy u, that is, all sum of returns;

(3)求解最优策略 (3) Solve the optimal strategy

s.t.f_j≥f_n stf _j ≥ f _n

(4)减少计算复杂度 (4) Reduce computational complexity

x_n＝T_n/b_nO x _n =T _n /b _n O

假设有N个用户竞争网络资源，第n个用户获得的资源为x_n，那么定义网络的公平系数为 Assuming that there are N users competing for network resources, and the resource obtained by the nth user is x _n , then the fairness coefficient of the network is defined as

$f f = = \frac{{((Σ Σ {x x}_{n no}))}^{22}}{NΣ NΣ {x x}_{n no}^{22}}$

下面结合具体参数和附图对本发明做进一步说明： Below in conjunction with concrete parameter and accompanying drawing, the present invention will be further described:

参数说明：WBAN仿真网络模型如图1所示，假设无线体域网中有三个传感器节点(节点1、2和3)和一个AP节点(节点4)。除了AP外，其他节点都有规律地采集人体生理参数，并将采集的数据通过共同的信道传送至AP。传输能量等级w＝{1,2,3}，对所有的传感器节点n，信道分布网络生存期用传输次数的期望来表示。 Parameter description: The WBAN simulation network model is shown in Figure 1. It is assumed that there are three sensor nodes (nodes 1, 2 and 3) and one AP node (node 4) in the wireless body area network. Except the AP, other nodes regularly collect physiological parameters of the human body, and transmit the collected data to the AP through a common channel. Transmission energy level w={1,2,3}, for all sensor nodes n, the channel distribution The network lifetime is expressed in terms of the expected number of transfers.

在图2中仿真了最优策略得到的网络生存期，并与其它几种传输协议做比较。这里，忽略了获取全局信道信息的开销，只关注于最优的极限性能和其它次优性能的差别。将最优策略和其他四种传输协议作比较：1)随机选取传感器节点传输；2)选取信道状况最好的传感器节点传输的机会调度协议；3)选取剩余能量最多的传感器节点传输的保守调度协议；4)DPLM协议，综合考虑信道状况信息和剩余能量信息来选取传感器节点传输数据。最大化网络生存期需要权衡好两个互相矛盾的目标：即最小化每个时隙的能量消耗和最小化网络死亡时的剩余能量。前者倾向于选择信道状况好的传感器节点来发送信息而后者倾向于选择剩余能量多的传感器节点来发送。上述的机会调度协议和保守调度协议只关注了两个互相矛盾的目标的其中一个，所以哪一个也不是最优的调度协议。DPLM协议综合考虑了信道状况信息和剩余能量信息，但其应用的是局部信道状况信息，即每个传感器节点只知道自己的信道状况信息和剩余能量信息，而不知道其他传感器节点的信息。虽然无法用马尔可夫决策过程建模，但DPLM仍然获得了近似最优的性能。应用最优策略的协议毫无疑问取得了最优的性能，并成为了其他协议对比的基准。 In Fig. 2, the network lifetime obtained by the optimal strategy is simulated, and compared with several other transmission protocols. Here, the overhead of obtaining global channel information is ignored, and only the difference between the optimal extreme performance and other suboptimal performances is focused. Comparing the optimal strategy with the other four transmission protocols: 1) Randomly select sensor nodes for transmission; 2) Opportunistic scheduling protocol for selecting sensor nodes with the best channel conditions for transmission; 3) Conservative scheduling for selecting sensor nodes with the most remaining energy for transmission protocol; 4) DPLM protocol, which comprehensively considers channel status information and residual energy information to select sensor nodes to transmit data. Maximizing network lifetime requires a trade-off between two conflicting goals: minimizing the energy consumption per slot and minimizing the remaining energy when the network dies. The former tends to choose sensor nodes with good channel conditions to send information, while the latter tends to choose sensor nodes with more residual energy to send information. The above opportunistic scheduling protocol and conservative scheduling protocol only focus on one of the two contradictory goals, so neither is the optimal scheduling protocol. The DPLM protocol considers the channel state information and the remaining energy information comprehensively, but it applies local channel state information, that is, each sensor node only knows its own channel state information and remaining energy information, and does not know the information of other sensor nodes. Although unable to be modeled with a Markov decision process, DPLM still achieves near-optimal performance. The protocol applying the optimal policy undoubtedly achieves the optimal performance and becomes the benchmark against which other protocols are compared. the

图2将网络生存期表示为初始能量的函数，随着传感器节点初始能量的增加，各种协议对应的生存期也相应的增大。应用最优策略的协议很好的显示了其最优的生存期性能，DPLM协议也获得了近似最优的性能。保守调度协议获得的网络生存期略好于机会调度协议，而随机选取传感器节点的协议的生存期性能最差。 Figure 2 shows the network lifetime as a function of the initial energy. With the increase of the initial energy of sensor nodes, the corresponding lifetimes of various protocols also increase accordingly. The protocol applying the optimal strategy shows its optimal lifetime performance well, and the DPLM protocol also obtains near-optimal performance. The network lifetime obtained by the conservative scheduling protocol is slightly better than that of the opportunistic scheduling protocol, while the lifetime performance of the protocol with randomly selected sensor nodes is the worst. the

图2是在没有公平性约束条件下的仿真，下面用带有约束的最优策略在网络对公平性要求不同的情况分别仿真，如图3所示。 Figure 2 is the simulation without fairness constraints. Next, the optimal strategy with constraints is used to simulate the network with different fairness requirements, as shown in Figure 3. the

仿真参数如下：假设体域网中有三个传感器节点，传输能量等级w＝{1,2,3}，对所有的传感器节点n，信道分布给定的公平系数阈值分别为f_n＝0.3,0.5,0.7,0.9，网络生存期用期望传输次数来表示。设每个传感器节点的权重相同，为b_n＝[13,13,13]。从图3中可以看出，网络的约束最优生存期随着对网络公平性要求的增加而减小。在对网络的公平性性能要求较高，如f_n＝0.9时的网络生存期降到很低。而对网络的公平性性能要求不高，如f_n＝0.3时的生存期性能依然很好。这说明在一定的范围内，公平性和网络的生存期是一对相互矛盾的性能指标。在实际中此方法能够根据应用场景在满足不同公平性的要求下达到约束最优的生存期。 The simulation parameters are as follows: Suppose there are three sensor nodes in the body area network, the transmission energy level w={1,2,3}, for all sensor nodes n, the channel distribution The given fairness coefficient thresholds are respectively f _n =0.3, 0.5, 0.7, and 0.9, and the network lifetime is represented by the expected number of transmissions. It is assumed that each sensor node has the same weight, which is b _n =[13,13,13]. It can be seen from Figure 3 that the constrained optimal lifetime of the network decreases as the requirement for network fairness increases. The requirement for the fairness performance of the network is relatively high, for example, when f _n =0.9, the lifetime of the network is reduced to a very low level. However, the fairness performance of the network is not highly required, for example, the lifetime performance is still very good when f _n =0.3. This shows that within a certain range, fairness and network lifetime are a pair of contradictory performance indicators. In practice, this method can achieve the constrained optimal lifetime according to the application scenarios while meeting different fairness requirements.

图4中设每个传感器节点的权重分别为b_n＝[0.7,0.2,0.1]，这表示第一个节点的重要性远远大于其它两个传感器节点。对比图3和图4可以看出，传感器权重值不同的时候，随着网络公平性要求的变化，生存期的变化比较大，而权重值相同的时候生存期的变化较小。这说明最优策略的方案对传感器权重值不同的情况作用更大，更加适用于对一个病人监测多项生理参数的情况。 In Fig. 4, the weights of each sensor node are set as b _n =[0.7, 0.2, 0.1] respectively, which means that the importance of the first node is far greater than that of the other two sensor nodes. Comparing Figure 3 and Figure 4, it can be seen that when the sensor weight values are different, the change in lifetime is relatively large with the change of network fairness requirements, while the change in lifetime is small when the weight values are the same. This shows that the optimal strategy scheme has a greater effect on the situation of different sensor weight values, and is more suitable for the situation of monitoring multiple physiological parameters for a patient.

Claims

1. A method for maximizing network lifetime transmission scheduling that satisfies fairness conditions, characterized in that it includes:

(1) Wireless body area network, which is a network that takes the human body as the monitoring object. Several sensor nodes with different functions are placed on the body surface or corresponding positions in the body to periodically monitor and record various physiological information. , each sensor node transmits the collected data to the aggregation node AP (Access Point) through Bluetooth, Zigbee, ultra-wideband or other human body communication technologies, and the AP transmits the information to the remote control center through the external network;

In the MAC layer protocol of the IEEE802.15.6 communication standard, time is divided into a superframe structure of equal length. The superframe structure includes the following four parts: control phase, contention access phase CAP, contention idle phase CFP, inactive phase, The CFP stage continues to be divided into several time slots. In the wireless body area network, time is the resource to be allocated to each sensor node. All sensor nodes in the wireless body area network share a channel, and there can only be one sensor node in each time slot. Transmit the collected data to the AP node, and the selected sensor node consumes the energy required to transmit the data;

(2) The problem of sensor node selection is modeled as a constrained Markov decision process. A transmission scheduling protocol corresponds to a policy u in the CMDP problem. The maximum network lifetime from state i is L ^* (i), Expressed as follows:

If the network uses a certain strategy u to get the most rewards before reaching the stop state (that is, the network has the longest lifetime), then we call this strategy the optimal strategy, denoted by u ^* ,

Define U _a ={u∈U:f≥f _n } as a set of feasible strategies. If the strategy u ^* ∈U _a satisfies the condition L(u ^* )≥L(u),(u∈U _a ), then u ^* is called the constrained optimal strategy;

The goal of CMDP is to find an optimal policy u ^* ∈ U _a to maximize the network lifetime L. The optimal transmission scheduling protocol is given by the optimal policy u ^* , that is, the optimal policy u ^* can be known in Which sensor node is selected for each time slot to send data;

The parameters in the CMDP model are introduced as follows:

(a) State space

In each time slot, the state i of the network is composed of the remaining energy e, the energy w required for transmission and the fairness factor f. We define the network state space S as follows:

S={i=(e,w,f)}

When the network lifetime is exhausted, the network reaches a special termination state S _t as follows:

e _n is the remaining energy of sensor node n, ε ₁ is the energy required for a time slot sensor node to transmit data with the minimum transmission power, e _n : e _n <ε ₁ means that the remaining energy of sensor node n is under any channel condition A transfer cannot be completed, e<w means a transfer failed;

(b) Action space

Use A to represent the set of actions. When the network is in the state i=(e,w,f)∈S, the action space can be expressed as:

A(i)=A[(e,w,f)]={n:e _n ≥ w _n }

According to the definition of terminal state, it can be concluded that the action space of any non-terminal state is non-empty;

(c) Transition probability

When the state of the network is i, after action a, the probability of the next state being j is

Where p(w')=Pr{W=w'} is the probability density function of W, which is determined by channel fading;

(d) Transfer Remuneration

After each data transfer, the network is rewarded with one unit until the network enters a stalled state, that is, if the state of the network is The immediate reward for this time slot is

(e) Constraints

Assuming that the network is in state i, if the fairness coefficient f of the network after action a is greater than or equal to a given threshold f _n , then we call action a a feasible action, and the formula can be expressed as follows:

f(i,a)≥f _n

f(i,a) represents the fairness coefficient of the network when it reaches the next state after the action of a in state i, and the feasible action set of the network in state i is represented by A _a (i);

(f) strategy

The policy u in the policy set U is a sequence u={u ₀ ,u ₁ ,...}, where u _n :S→{1,...,N} represents the sensor node selected in the nth time slot , u _n (i ₁ ,a ₁ ,i ₂ ,a ₂ ,...,i _n-1 ,a _n-1 ,i _n ) is the conditional probability measure of A, measured at the moment when the energy of the first node is exhausted Therefore, the lifetime L of the entire network can be described by the sum of all rewards before the network reaches the stop state S _t . Define L _u (i) as the lifetime of the network starting from state i and using strategy u, that is, all sum of returns;

(3) Solve the optimal strategy

Starting from state i, the maximum network lifetime L ^* (i) that satisfies the constraints is the only solution to the following Bellman optimal equation,

stf _j ≥ f _n

Where f _j is the fairness coefficient of the network in state j, and f _n is the fairness threshold given according to different application scenarios;

In fact, the above formula can also be written as

The optimal policy u of the transmission scheduling scheme can be obtained by the following formula

(4) Reduce computational complexity

An equivalent Bellman optimality equation is written as follows:

Therefore, the constrained optimal strategy can also be obtained by the following formula

In the implementation process, at the beginning of each time slot, the AP node broadcasts a beacon signal to wake up all sensor nodes. In order for the AP node to obtain global channel information, all sensor nodes need to send a pilot signal to reply to the AP’s The beacon signal, the AP node uses the received signal to estimate the channel conditions of all nodes and obtain the energy w required for transmission, and then calculate the optimal strategy u according to the state (e, w, f), and define the optimal strategy, namely Knowing the sensor node that should be selected in each time slot, finally, the AP node broadcasts the ID number of the selected sensor node, and the sensor node starts to transmit the data it collects. Since the AP node knows the channel status of all sensor nodes, it can Continue to update the network status to prepare for the arrival of the next time slot. the

2. A kind of maximizing network lifetime transmission scheduling method satisfying the fairness condition according to claim 1, characterized in that, the fairness coefficient f is defined as follows: define (T ₁ , T ₂ ,..., T _N ) represents the actual number of transmissions of each sensor node, O represents the total number of transmissions, (b ₁ , b ₂ ,...,b _N ) is the weight representing the importance of each sensor node. For sensor node n, the normalized number of transmissions is defined as

x _n =T _n /b _n O

Assuming that there are N users competing for network resources, and the resource obtained by the nth user is x _n , then the fairness coefficient of the defined network is

f takes values between 0 and 1. If the fairness coefficient f=1 of the network, it means that the network is completely fair. The larger the fairness coefficient of the network, the higher the fairness, and vice versa. the