WO2024066284A1 - 一种与源址无关的分布式状态监测方法 - Google Patents
一种与源址无关的分布式状态监测方法 Download PDFInfo
- Publication number
- WO2024066284A1 WO2024066284A1 PCT/CN2023/086658 CN2023086658W WO2024066284A1 WO 2024066284 A1 WO2024066284 A1 WO 2024066284A1 CN 2023086658 W CN2023086658 W CN 2023086658W WO 2024066284 A1 WO2024066284 A1 WO 2024066284A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- time slot
- sensor
- state
- index
- state variable
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000012544 monitoring process Methods 0.000 title claims abstract description 29
- 230000005540 biological transmission Effects 0.000 claims abstract description 35
- 230000004927 fusion Effects 0.000 claims abstract description 35
- 230000004913 activation Effects 0.000 claims abstract description 26
- 230000008569 process Effects 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000001514 detection method Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 101001121408 Homo sapiens L-amino-acid oxidase Proteins 0.000 description 1
- 102100026388 L-amino-acid oxidase Human genes 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Classifications
-
- 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
- H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/08—Testing, supervising or monitoring using real traffic
Definitions
- the present invention relates to the field of wireless communications, and in particular to a distributed state monitoring method that is independent of a source address.
- IoT applications there are various IoT applications in IoT scenarios, one of which is distributed state monitoring.
- distributed state monitoring task multiple sensors observe the output of the system at different locations and send the observation results to the fusion center to estimate the global system state vector.
- each sensor can only obtain observations of some state variables in the global state vector, and the fusion center only focuses on the state information embedded in the observation rather than which sensor sent it.
- Due to the passivity of sensor transmission uplink transmission can be performed based on a random access protocol that is independent of the source address.
- a sensor can only obtain observations of some state variables and only a small number of sensors are activated at the same time, the observation of the global state vector through only one time slot transmission is insufficient.
- the observation of the global state vector will be inaccurate.
- the insufficiency and inaccuracy of the observation of state variables require observations in multiple time slots to obtain accurate observations of the global state vector of the system.
- the user in order to solve the problem of too high a dimension of the common codebook, the user first divides the information sequence into several sub-blocks, and then adds check bits to establish a check relationship between different information sub-blocks. At the receiving end, all transmitted information sub-blocks are first detected, and then a tree decoder is used to decode and splice to restore the original information sequence.
- the purpose of the present invention is to propose an efficient source-independent distributed state monitoring method for distributed state monitoring tasks in the Internet of Things scenario.
- a source-independent distributed status monitoring method characterized by comprising the following steps:
- the fusion center detects and estimates the information sub-blocks corresponding to the codewords transmitted by the sensor and the corresponding superposition channel gain values from the received signal;
- the fusion center obtains an estimated value of each state variable and a reliability index of the estimated value according to a number of observation values of each state variable observed by multiple sensors and a channel gain of each observation value; when the reliability index of the state variable is greater than a set reliability threshold, it indicates that the observation of the state variable is reliable;
- the entire monitoring area is divided into different areas.
- the fusion center broadcasts the reliability index of each state variable estimation value. If the estimation of all observable state variables in a region is reliable, then the state variable estimation of this region is completed; the fusion center instructs the sensors in the area where the state estimation is not completed to activate in the next time slot. All sensors determine the activation probability in the next transmission process according to the reliability index of the observable state variables;
- step S5 Continue to execute the process from step S1 to step S4 until the fusion center obtains reliable estimates of all state variables of the monitoring area system.
- step S1 The encoding method described in step S1 is:
- Set up codebook in Represents the complex field, each column of A represents a codeword, and the codewords have The length of each codeword is L c ; the whole system has a total of N o states, and the jth state vector is The length is bs bits; the index vector of this state is represented by Indicates that the length is b I bits; information sub-block The length is N m b I + b s bits.
- a time slot is divided into T sub-time slots; for the t-th sub-time slot, the k-th activated sensor maps the N m- bit information sub-block to be sent into a value range of 1 to Integer The kth activated sensor converts the The codewords represented by the columns are sent to the fusion center.
- step S2 The detection and estimation method described in step S2 is:
- the vector X is recovered from the received signal Y using the compressed sensing method to obtain the codeword transmitted by the sensor and the corresponding channel gain amplitude set; then, according to the codeword index, it is converted into a binary vector, which is the information sub-block sent by the user; in the t-th sub-time slot within the l-th time slot, for the n-th codeword, it is converted into a binary vector to obtain the information sub-block
- the channel gain amplitude corresponding to this codeword is Where
- the detected information sub-block set is The superposition channel gain set is
- the reliability index calculation method described in step S3 is:
- the data in T sub-time slots can be obtained. and from An element in can obtain the bth observation value of the jth state variable
- the subscript (l) indicates the lth time slot; if the jth state variable is observed by different sensors, the different channel gain amplitudes corresponding to the same observation value are superimposed to obtain the corresponding superimposed channel gain amplitude Finally, after the observation of the lth time slot, all the observed values of the jth state variable constitute an observation value set
- the superposition channel gain amplitude corresponding to each observation value constitutes a superposition channel gain amplitude set Where
- the estimation of the j-th state variable is regarded as a classification problem; all elements in the superimposed channel gain amplitude set are concatenated into a vector as the input of the softmax function, and then the index of the estimated value of the j-th state variable is output.
- the domain of x is 0 ⁇ x ⁇ 1, and the parameter p is the scaling factor; For each g, Take The maximum g; the estimated value can be expressed as make Represents the reliability index of the estimated value of the j-th state variable, and sets the threshold of the number of observations and the reliability threshold;
- c is the threshold of the number of observations, represents the channel gain amplitude corresponding to the qj -th observation value of the j-th state variable;
- ⁇ is the reliability threshold, then Indicates that the observation of the jth state is reliable.
- step S4 The process described in step S4 is:
- the activation probability of the kth sensor in the l+1th time slot is It can be expressed as Where Ok represents the index set of state variables that the kth sensor can observe, that is Denotes O k is a subset of ⁇ 1,2,...,N o ⁇ ;
- O k is a subset of ⁇ 1,2,...,N o ⁇ ;
- the contribution weight of the reliability index of the jth state variable to the activation probability of the kth sensor in the l+1th time slot is represented by
- the entire area to be observed is divided into Na areas, and the state index set that the sensor can observe in the nth area is The state variables are estimated in a region. If the estimates of all observable state variables in a region are reliable, then the state variable estimation of this region can be considered to be completed.
- the fusion center only needs to instruct the sensors distributed in other areas to activate in the next time slot; in the transmission of the next time slot, it is hoped that as few areas as possible will be activated so that all the state variables to be observed can be covered; this problem is modeled as a set coverage optimization problem, as shown below:
- n a region is selected each time, where n a can be expressed as Represents the state index set that the sensor can observe in the n a region; then the index n a is removed from the set Remove and add to collection
- the collection Indicates the index set of the area that needs to be activated in the l+1th time slot; repeat the above selection process until in Indicates that all n a belong to U represents all Take the union; finally The index of the area that needs to be activated in the l+1th time slot will be included;
- the activation probability of a sensor is represents the sensor index set in the nth region, where p 0 represents the activation probability of the sensor in the first time slot, and the specific value should be set according to the situation of the scene; represents the activation probability of the kth sensor in the l+1th time slot, according to It is obtained that the sensors in other areas except the nth area remain dormant in the next transmission time slot.
- the distributed state observation method proposed by the invention is independent of the source address, and adopts a scheme of multiple time slot transmission, which solves the problem that one time slot transmission makes the observation of the global state vector insufficient and inaccurate.
- the existing method uses a large number of check bits, has a low code rate, and is inefficient in completing the distributed state monitoring task.
- This scheme realizes the use of a small number of time slot transmissions to efficiently complete the accurate observation of the global state vector of the system.
- FIG1 is a schematic diagram of a distributed status monitoring scenario provided by an embodiment of the present invention.
- 2 is a graph showing the relationship between the average number of transmission time slots and the signal-to-noise ratio when the distributed state monitoring method provided by an embodiment of the present invention is compared with the baseline method under different values of the number of state variables;
- 3 is a graph showing the relationship between the minimum mean square error of state variable estimation and the maximum allowed number of transmission time slots when the distributed state monitoring method provided by an embodiment of the present invention is compared with the baseline method when the number of state variables is at different values.
- the scenario diagram of distributed state monitoring is shown in Figure 1.
- the fusion center processes the received signal, estimates the global state vector of the system, and broadcasts the estimated reliability index of each state variable to all sensors.
- the sensor determines the activation probability in the next transmission time slot based on the estimated reliability index of the observable state variable. The transmission process continues until the fusion center obtains reliable estimates of all state variables of the system.
- This embodiment provides a source-independent distributed status monitoring method, which includes the following steps:
- the observation results and status of each state are combined.
- the state index is spliced into information sub-blocks, which are encoded and sent to the fusion center through wireless channels.
- the encoding method is:
- Set up codebook in Represents the complex field, each column of A represents a codeword, and the codewords have The length of each codeword is L c ; the whole system has a total of N o states, and the jth state vector is The length is bs bits; the index vector of this state is represented by Indicates that the length is b I bits; information sub-block The length is N m b I + b s bits; a time slot is divided into T sub-time slots; for the t-th sub-time slot, the k-th activated sensor, the N m -bit information sub-block to be sent is mapped into a value range of 1 to Integer The kth activated sensor converts the The codewords represented by the columns are sent to the fusion center.
- the fusion center detects and estimates the information sub-blocks and corresponding superimposed channel gain values corresponding to the codewords transmitted by the sensor from the received signal.
- the detection and estimation method is:
- the vector X is recovered from the received signal Y using the compressed sensing method to obtain the codeword transmitted by the sensor and the corresponding channel gain amplitude set; then, according to the codeword index, it is converted into a binary vector, which is the information sub-block sent by the user; in the t-th sub-time slot within the l-th time slot, for the n-th codeword, it is converted into a binary vector to obtain the information sub-block
- the channel gain amplitude corresponding to this codeword is Where
- the detected information sub-block set is The superposition channel gain set is
- the fusion center obtains the estimated value of each state variable and the reliability index of the estimated value according to several observed values of each state variable and the channel gain of each observed value.
- the reliability index is calculated as follows:
- the data in T sub-time slots can be obtained. and from An element in can obtain the bth observation value of the jth state variable
- the subscript (l) indicates the lth time slot. If the jth state variable is observed by different sensors, the different channel gain amplitudes corresponding to the same observation value are superimposed to obtain the corresponding Superposition channel gain amplitude Finally, after the observation of the lth time slot, all the observed values of the jth state variable constitute an observation value set
- the superposition channel gain amplitude corresponding to each observation value constitutes a superposition channel gain amplitude set Where
- the estimation of the j-th state variable is regarded as a classification problem; all elements in the superimposed channel gain amplitude set are concatenated into a vector as the input of the softmax function, and then the index of the estimated value of the j-th state variable is output.
- the domain of x is 0 ⁇ x ⁇ 1, and the parameter p is the scaling factor; For each g, Take The maximum g; the estimated value can be expressed as make Represents the reliability index of the estimated value of the j-th state variable, and sets the threshold of the number of observations and the reliability threshold;
- c is the threshold of the number of observations, represents the channel gain amplitude corresponding to the qj -th observation value of the j-th state variable;
- ⁇ is the reliability threshold, then Indicates that the observation of the jth state is reliable.
- the fusion center broadcasts the reliability index of each state variable estimate, and all sensors decide the activation probability in the next transmission process based on the reliability index of the observable state variables.
- the sensor determines the activation probability by:
- the activation probability of the kth sensor in the l+1th time slot is It can be expressed as Where Ok represents the index set of state variables that the kth sensor can observe, that is Denotes O k is a subset of ⁇ 1,2,...,N o ⁇ ;
- O k is a subset of ⁇ 1,2,...,N o ⁇ ;
- the contribution weight of the reliability index of the jth state variable to the activation probability of the kth sensor in the l+1th time slot is represented by
- the entire area to be observed is divided into Na areas, and the state index set that the sensor can observe in the nth area is The state variables are estimated in a region. If the estimates of all observable state variables in a region are reliable, then the state variable estimation of this region can be considered to be completed.
- the fusion center only needs to instruct the sensors distributed in other areas to activate in the next time slot; in the transmission of the next time slot, it is hoped that as few areas as possible will be activated so that all the state variables to be observed can be covered; this problem is modeled as a set coverage optimization problem, as shown below:
- n a region is selected each time, where n a can be expressed as Represents the state index set that the sensor can observe in the n a region; then the index n a is removed from the set Remove and add to collection
- the collection Indicates the index set of the area that needs to be activated in the l+1th time slot; repeat the above selection process until in Indicates that all n a belong to U represents all Take the union; finally The index of the area that needs to be activated in the l+1th time slot will be included;
- the activation probability of a sensor is represents the sensor index set in the nth region, where p 0 represents the activation probability of the sensor in the first time slot, and the specific value should be set according to the situation of the scene; represents the activation probability of the kth sensor in the l+1th time slot, according to It is obtained that the sensors in other areas except the nth area remain dormant in the next transmission time slot.
- step 1) Continue to execute the process from step 1) to step 4) until the fusion center obtains reliable estimates of all state variables of the system.
- the distributed state monitoring scheme of the present invention is compared with the random activation scheme of the baseline. Under the same signal-to-noise ratio conditions, the average number of transmission time slots required to complete the observation of the global state of the system is significantly reduced.
- Figure 3 shows that the distributed state monitoring scheme proposed by the present invention can significantly reduce the minimum mean square error of state variable estimation after a small number of time slot transmissions compared to the random activation scheme of the baseline.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
本发明公开了一种与源址无关的分布式状态监测方法。本发明基于与源址无关的随机接入协议，所有传感器共享一个公共码本，融合中心只需恢复传输的消息序列而不关心活跃传感器的身份信息。首先，激活传感器观测部分状态变量并将观测结果上行传输至融合中心，融合中心对观测信息进行处理、估计全局状态向量并广播关于状态变量估计的可靠性指标，然后传感器根据可观测状态变量估计的可靠性指标决定在下一个时隙内的激活概率，直到对系统中所有状态变量的估计都可靠则任务完成。本方案实现了利用少量时隙传输，高效地完成对系统全局状态向量的准确观测。
Description
本发明涉及无线通信领域，尤其涉及一种与源址无关的分布式状态监测方法。
随着物联网的快速发展，物联网设备的数量有了大幅增长。大量的物联网设备可以支持多样的物联网应用，关键是要提供快速、可靠的接入。传统的免授权随机接入技术不适用于物联网场景。这是由于在免授权随机接入方案中，每个潜在用户都被分配了一个特定的导频序列且潜在用户数量庞大，这样的导频开销是不可接受的。为了解决这一问题，一种与源址无关的随机接入技术被提出。在该方案中，所有潜在用户共享一个公共码本，基站端只需恢复传输的消息序列而不关心活跃用户的身份。由于用户无需传输导频序列，该方案可以避免高昂的导频开销，满足了物联网场景的关键需求。
在物联网场景下有多样的物联网应用，其中一个应用是分布式状态监测。在分布式状态监测任务中，多个传感器在不同位置观测系统的输出并将观测结果发送至融合中心，估计全局的系统状态向量。对于一个典型的线性系统，每个传感器只能获取对全局状态向量中部分状态变量的观测，融合中心只关注于嵌入在观测中的状态信息而不是哪个传感器发送的。由于传感器传输的无源性，可以以基于与源址无关的随机接入协议进行上行传输。然而，由于一个传感器只能获取对部分状态变量的观测且同一时刻只有少量传感器激活，仅通过一个时隙的传输对全局状态向量的观测存在不充分性。另外，由于检测过程可能存在漏检、误检，会导致对全局状态向量观测的不准确性。对状态变量观测的不充分性和不准确性使得需要多个时隙的观测才能获取对系统全局状态向量的准确观测。然而，在现有的与源址无关的随机接入协议中，为了解决公共码本维度过高的问题，用户先将信息序列分为若干个子块，然后添加校验比特建立不同信息子块间的校验关系。在接收端先检测所有传输的信息子块，然后使用树形译码器进行译码、拼接恢复出原始的信息序列。这样做需要大量的校验比特保证译码的准确性，使得传输码率较低。由于传输码率较低且需要多个时隙的观测，使得完成分布式状态监测任务的效率低下。因此，如何设计一种方案使得传感器能快速、准确地完成对系统全局状态向量的估计成为了关键性问题。
发明内容
本发明的目的是为物联网场景下的分布式状态监测任务提出了一种高效的与源址无关的分布式状态监测方法。
本发明所采用的具体技术方案如下：
一种与源址无关的分布式状态监测方法，其特征在于包括如下步骤：
S1、激活传感器，获取对整个监测区系统状态的观测并生成状态索引，把获取到的每个状态的观测结果和状态索引拼接成信息子块，将信息子块进行编码后通过无线信道发送至融合中心；
S2、融合中心从接收信号中检测、估计出传感器传输的码字对应的信息子块和相应的叠加信道增益值；
S3、融合中心根据每个状态变量由多个传感器观测的若干观测值和每个观测值的信道增益，得到该状态变量的估计值以及该估计值的可靠性指标；当该状态变量的可靠性指标大于设定的可靠性阈值时，则表明对该状态变量的观测是可靠的；
S4、将整个监测区分为不同的区域，融合中心广播每个状态变量估计值的可靠性指标，如果一个区域内对所有可观测状态变量的估计都是可靠的，那么对这个区域的状态变量估计已完成；融合中心指示那些未完成状态估计区域内的传感器在下一个时隙内激活，所有传感器根据可观测状态变量的可靠性指标决定在下一次传输过程中的激活概率；
S5、不断执行步骤S1至步骤S4的过程直到融合中心获取了对监测区系统所有状态变量的可靠估计。
步骤S1中所述的编码方法为：
设置码本其中代表复数域，A的每一列都表示一个码字，码字共有个，每个码字的长度为Lc；整个系统共有No个状态，第j个状态向量用表示，长度为bs比特；该状态的索引向量用表示，长度为bI比特；信息子块长度为Nm＝bI+bs比特。将一个时隙分为T个子时隙；对于其中的第t个子时隙，第k个激活传感器，将要发送的Nm比特信息子块映射成取值范围为1到的整数第k个激活传感器将码本A中的第列表示的码字发送至融合中心。
步骤S2中所述的检测、估计方法为：
在物联网场景中，共有Ktotal个传感器，一次传输过程中仅Ka个传感器激活，即Ka＜＜Ktotal；接收信号可以表示为Y＝AΔH+Z＝AX+Z，其中表示所有传感器的信道；表示选择矩阵，Δ矩阵中的元素δn,k表示第k个传感器是否传输了第n个码字；Z为噪声且服从均值为0的复高斯分布；向量的每一个非零元素表示传输这个码字的传感器到融合中心的信道增益；
使用压缩感知方法从接收信号Y中恢复出向量X，得到传感器传输的码字和相应的信道增益幅值集合；然后根据码字索引，将其转化为二进制向量，此二进制向量为用户发送的信息子块；在第l个时隙内的第t个子时隙，对于第n个码字，将其转化为二进制向量得到信息子块该码字相应的信道增益幅值为其中|·|表示取幅值，角标(l,t)表示第l个时隙内的第t个子时隙；令检测到的码字索引集合为检测到的信息子块集合为叠加信道增益集合为
步骤S3中所述的可靠性指标计算方法为：
在第l个时隙内，能够获得T个子时隙内的和从中的一个元素能够获得对第j个状态变量的第b个观测值其中角标(l)表示第l个时隙；若第j个状态变量被不同的传感器观测到，将同一个观测值对应的不同信道增益幅值叠加得到相应的叠加信道增益幅值最终经过第l个时隙的观测，对于第j个状态变量的所有观测值构成一个观测值集合每个观测值对应的叠加信道增益幅值构成叠加信道增益幅值集合其中|·|c表示一个集合的元素个数；
将对第j个状态变量的估计视为一个分类问题；将叠加信道增益幅值集合中的所有元素拼接为一个向量后作为softmax函数的输入，然后输出得到对第j个状态变量估计值的索引表示为
其中f(x)为sigmoid函数的截断和放缩，可以表示为
其中x的定义域为0≤x≤1，参数p为放缩因子；表示对每一个g对应的取使最大的g；估计值可以表示为令表示对第j个状态变量估计值的可靠性指标，并设定观测次数的阈值和可靠性阈值；
可以表示为
其中表示经过第l时隙的传输后，获得对第j个状态的观测次数，c为观测次数的阈值，
表示第j个状态变量的第qj个观测值所对应的信道增益幅值；∈为可靠性阈值，则表示对第j个状态的观测是可靠的。
步骤S4中所述的过程为：
令则第k个传感器在第l+1个时隙内的激活概率可以表示为其中Ok表示第k个传感器能够观测到的状态变量索引集合，即表示Ok是{1,2,...,No}的子集；表示第j个状态变量的可靠性指标对第k个传感器在第l+1个时隙内激活概率的贡献权重
整个待观测区域被分为Na个区域，第na个区域内传感器能观测到的状态索引集合为以一个区域为单位进行状态变量估计，如果一个区域内对所有可观测状态变量的估计都是可靠的，那么可以认为对这个区域的状态变量估计已完成；
融合中心只需指示那些分布在其他区域内的传感器在下一个时隙内激活；在下一个时隙的传输中，希望尽可能少的区域激活就能使得所有待观测的状态变量被覆盖；将这个问题建模为一个集合覆盖的最佳化问题，如下所示：
矩阵表示不同区域内能观测到的状态变量；为矩阵B的元素，表示第na个区域内的传感器是否能获取对第j个状态变量的观测；表示选择第na个区域的代价，的值被设置为1；表示是否选择第na个区域；表示至少一个激活区域内的传感器获取对第j个状态变量的观测；
采用贪婪算法求解这个集合覆盖的最佳化问题，令表示一个集合，该集合内的元素为在第l+1个时隙内需要继续观测的状态变量的索引；令表示一个集合，该集合内的元素为含有经过第l个时隙观测后不可靠状态变量估计区域的索引；在集合内，每次选取第na个区域,其中na可以表示为表示第na个区域内传感器能观测到的状态索引集合；然后将索引na从集合中移除并加入集合其中集合
表示在第l+1个时隙内需要激活区域的索引集合；重复上述选择过程直到其中表示所有的na属于U·表示对所有的取并集；最终将包含在第l+1个时隙内需要激活区域的索引；
在一个传输时隙内，激活传感器的数量将被控制；分布在第个区域内的第个传感器的激活概率为表示第na个区域内的传感器索引集合，其中p0表示在第一个时隙内传感器的激活概率，具体数值要根据场景的情况设定；表示第l+1个时隙内第k个传感器的激活概率，根据得到；除了第na个区域之外的其他区域内的传感器在下一个传输时隙内保持休眠。
本发明具有的有益效果是：本发明提出的与源址无关的分布式状态观测方法，采用多个时隙传输的方案，解决了一个时隙传输使得对全局状态向量观测的不充分和不准确的问题。另外，还解决了现有方法使用大量校验比特，码率较低，完成分布式状态监测任务效率低下的问题。本方案实现了利用少量时隙传输，高效地完成对系统全局状态向量的准确观测。
图1是本发明实施例提供的分布式状态监测的场景示意图；
图2是本发明实施例提供的状态变量数处于不同值下分布式状态监测方法与基线方法比较时，平均传输时隙数与信噪比的关系图；
图3是本发明实施例提供的状态变量数处于不同值下分布式状态监测方法与基线方法比较时，状态变量估计的最小均方误差与最大允许传输时隙数的关系图。
以下结合附图对本发明具体实施方式作进一步详细说明。
本实施例中，分布式状态监测的场景图如图1所示。场景中有一个融合中心，分布着多个传感器。每个时隙内，仅有少量传感器激活，获取对系统状态向量的观测，并以与源址无关的随机接入协议将观测结果上传至融合中心。融合中心对接收信号进行处理、估计系统的全局状态向量，并将每个状态变量估计的可靠性指标广播给所有传感器。传感器根据可观测状态变量估计的可靠性指标决定在下一个传输时隙内的激活概率。传输过程持续直到融合中心获取了对系统所有状态变量的可靠估计。
本实施例提供了一种与源址无关的分布式状态监测方法，其包括如下步骤：
1)在物联网场景中，激活设备获取对系统状态的观测后，把对每个状态的观测结果和状
态索引拼接成信息子块，将信息子块进行编码后通过无线信道发送至融合中心。
在本步骤中，编码方法为：
设置码本其中代表复数域，A的每一列都表示一个码字，码字共有个，每个码字的长度为Lc；整个系统共有No个状态，第j个状态向量用表示，长度为bs比特；该状态的索引向量用表示，长度为bI比特；信息子块长度为Nm＝bI+bs比特；将一个时隙分为T个子时隙；对于其中的第t个子时隙，第k个激活传感器，将要发送的Nm比特信息子块映射成取值范围为1到的整数第k个激活传感器将码本A中的第列表示的码字发送至融合中心。
2)融合中心从接收信号中检测、估计出传感器传输的码字对应的信息子块和相应的叠加信道增益值。
在本步骤中，检测、估计方法为：
在物联网场景中，共有Ktotal个传感器，一次传输过程中仅Ka个传感器激活，即Ka＜＜Ktotal；接收信号可以表示为Y＝AΔH+Z＝AX+Z，其中表示所有传感器的信道；表示选择矩阵，Δ矩阵中的元素δn,k表示第k个传感器是否传输了第n个码字；Z为噪声且服从均值为0的复高斯分布；向量的每一个非零元素表示传输这个码字的传感器到融合中心的信道增益；
使用压缩感知方法从接收信号Y中恢复出向量X，得到传感器传输的码字和相应的信道增益幅值集合；然后根据码字索引，将其转化为二进制向量，此二进制向量为用户发送的信息子块；在第l个时隙内的第t个子时隙，对于第n个码字，将其转化为二进制向量得到信息子块该码字相应的信道增益幅值为其中|·|表示取幅值，角标(l,t)表示第l个时隙内的第t个子时隙；令检测到的码字索引集合为检测到的信息子块集合为叠加信道增益集合为
3)融合中心根据每个状态变量的若干个观测值和每个观测值的信道增益得到该状态变量的估计值以及该估计值的可靠性指标。
在本步骤中，可靠性指标计算方法为：
在第l个时隙内，能够获得T个子时隙内的和从中的一个元素能够获得对第j个状态变量的第b个观测值其中角标(l)表示第l个时隙；若第j个状态变量被不同的传感器观测到，将同一个观测值对应的不同信道增益幅值叠加得到相应的
叠加信道增益幅值最终经过第l个时隙的观测，对于第j个状态变量的所有观测值构成一个观测值集合每个观测值对应的叠加信道增益幅值构成叠加信道增益幅值集合其中|·|c表示一个集合的元素个数；
将对第j个状态变量的估计视为一个分类问题；将叠加信道增益幅值集合中的所有元素拼接为一个向量后作为softmax函数的输入，然后输出得到对第j个状态变量估计值的索引表示为
其中f(x)为sigmoid函数的截断和放缩，可以表示为
其中x的定义域为0≤x≤1，参数p为放缩因子；表示对每一个g对应的取使最大的g；估计值可以表示为令表示对第j个状态变量估计值的可靠性指标，并设定观测次数的阈值和可靠性阈值；
可以表示为
其中表示经过第l时隙的传输后，获得对第j个状态的观测次数，c为观测次数的阈值，表示第j个状态变量的第qj个观测值所对应的信道增益幅值；∈为可靠性阈值，则表示对第j个状态的观测是可靠的。
4)融合中心广播每个状态变量估计值的可靠性指标，所有传感器根据可观测状态变量的可靠性指标决定在下一次传输过程中的激活概率。
在本步骤中，传感器决定激活概率的方法为：
令则第k个传感器在第l+1个时隙内的激活概率可以表示为其中Ok表示第k个传感器能够观测到的状态变量索引集合，即表示Ok是{1,2,...,No}的子集；表示第j个状态变量的可靠性指标对第k个传感器在第l+1个时隙内激活概率的贡献权重
整个待观测区域被分为Na个区域，第na个区域内传感器能观测到的状态索引集合为以一个区域为单位进行状态变量估计，如果一个区域内对所有可观测状态变量的估计都是可靠的，那么可以认为对这个区域的状态变量估计已完成；
融合中心只需指示那些分布在其他区域内的传感器在下一个时隙内激活；在下一个时隙的传输中，希望尽可能少的区域激活就能使得所有待观测的状态变量被覆盖；将这个问题建模为一个集合覆盖的最佳化问题，如下所示：
矩阵表示不同区域内能观测到的状态变量；为矩阵B的元素，bj,na表示第na个区域内的传感器是否能获取对第j个状态变量的观测；表示选择第na个区域的代价，的值被设置为1；表示是否选择第na个区域；表示至少一个激活区域内的传感器获取对第j个状态变量的观测；
采用贪婪算法求解这个集合覆盖的最佳化问题，令表示一个集合，该集合内的元素为在第l+1个时隙内需要继续观测的状态变量的索引；令表示一个集合，该集合内的元素为含有经过第l个时隙观测后不可靠状态变量估计区域的索引；在集合内，每次选取第na个区域,其中na可以表示为表示第na个区域内传感器能观测到的状态索引集合；然后将索引na从集合中移除并加入集合其中集合表示在第l+1个时隙内需要激活区域的索引集合；重复上述选择过程直到其中表示所有的na属于U·表示对所有的取并集；最终将包含在第l+1个时隙内需要激活区域的索引；
在一个传输时隙内，激活传感器的数量将被控制；分布在第个区域内的第个传感器的激活概率为表示第na个区域内的传感器索引集合，其中p0表示在第一个时隙内传感器的激活概率，具体数值要根据场景的情况设定；
表示第l+1个时隙内第k个传感器的激活概率，根据得到；除了第na个区域之外的其他区域内的传感器在下一个传输时隙内保持休眠。
5)不断执行步骤1)至步骤4)的过程直到融合中心获取了对系统所有状态变量的可靠估计。
通过计算机仿真可以看出：如图2所示，本发明的分布式状态监测方案相较于基线的随机激活方案，在相同的信噪比条件下，完成对系统全局状态观测所需的平均传输时隙数有明显的下降。图3表明本发明提出的分布式状态监测方案，相较于基线的随机激活方案，可以使得经过少量的时隙传输，状态变量估计的最小均方误差有明显下降。这些优势主要是是因为本文方案利用了每个时隙对全局状态向量估计的结果指导传感器在下一个时隙内的传输。因此，本发明提出的与源址无关的分布式状态监测方案提供了一种高效的系统全局状态向量估计方法。
上述实施例用来解释说明本发明，而不是对本发明进行限制，在本发明的精神和权利要求的保护范围内，对本发明作出的任何修改和改变，都落入本发明的保护范围。
Claims (5)
- 一种与源址无关的分布式状态监测方法，其特征在于，包括如下步骤：S1、激活传感器，获取对整个监测区系统状态的观测并生成状态索引，把获取到的每个状态的观测结果和状态索引拼接成信息子块，将信息子块进行编码后通过无线信道发送至融合中心；S2、融合中心从接收信号中检测、估计出传感器传输的码字对应的信息子块和相应的叠加信道增益值；S3、融合中心根据每个状态变量由多个传感器观测的若干观测值和每个观测值的信道增益，得到该状态变量的估计值以及该估计值的可靠性指标；当该状态变量的可靠性指标大于设定的可靠性阈值时，则表明对该状态变量的观测是可靠的；S4、将整个监测区分为不同的区域，融合中心广播每个状态变量估计值的可靠性指标，如果一个区域内对所有可观测状态变量的估计都是可靠的，那么对这个区域的状态变量估计已完成；融合中心指示那些未完成状态估计区域内的传感器在下一个时隙内激活，所有传感器根据可观测状态变量的可靠性指标决定在下一次传输过程中的激活概率；S5、不断执行步骤S1至步骤S4的过程直到融合中心获取了对监测区系统所有状态变量的可靠估计。
- 根据权利要求1所述的一种与源址无关的分布式状态监测方法，其特征在于，所述步骤S1中的编码方法具体步骤为：S1.1、设置码本其中代表复数域，A的每一列都表示一个码字，码字共有个，每个码字的长度为Lc；整个系统共有No个状态，第j个状态向量用表示，长度为bs比特；该状态的索引向量用表示，长度为bI比特；信息子块长度为Nm＝bI+bs比特；S1.2、将一个时隙分为T个子时隙；对于其中的第t个子时隙，第k个激活传感器，将要发送的Nm比特信息子块映射成取值范围为1到的整数S1.3、第k个激活传感器将码本A中的第列表示的码字发送至融合中心。
- 根据权利要求2所述的一种与源址无关的分布式状态监测方法，其特征在于，所述步骤S2中所述的检测、估计方法为：在物联网场景中，共有Ktotal个传感器，一次传输过程中仅Ka个传感器激活，即Ka＜＜Ktotal；接收信号可以表示为Y＝AΔH+Z＝AX+Z，其中表示所有传感器的 信道；表示选择矩阵，Δ矩阵中的元素δn,k表示第k个传感器是否传输了第n个码字；Z为噪声且服从均值为0的复高斯分布；向量的每一个非零元素表示传输这个码字的传感器到融合中心的信道增益；使用压缩感知方法从接收信号Y中恢复出向量X，得到传感器传输的码字和相应的信道增益幅值集合；然后根据码字索引，将其转化为二进制向量，此二进制向量为用户发送的信息子块；在第l个时隙内的第t个子时隙，对于第n个码字，将其转化为二进制向量得到信息子块该码字相应的信道增益幅值为其中|·|表示取幅值，角标(l,t)表示第l个时隙内的第t个子时隙；令检测到的码字索引集合为检测到的信息子块集合为叠加信道增益集合为
- 根据权利要求3所述的一种与源址无关的分布式状态监测方法，其特征在于，所述步骤S3中所述的可靠性指标计算方法为：在第l个时隙内，能够获得T个子时隙内的和从中的一个元素能够获得对第j个状态变量的第b个观测值其中角标(l)表示第l个时隙；若第j个状态变量被不同的传感器观测到，将同一个观测值对应的不同信道增益幅值叠加得到相应的叠加信道增益幅值最终经过第l个时隙的观测，对于第j个状态变量的所有观测值构成一个观测值集合每个观测值对应的叠加信道增益幅值构成叠加信道增益幅值集合其中|·|c表示一个集合的元素个数；将对第j个状态变量的估计视为一个分类问题；将叠加信道增益幅值集合中的所有元素拼接为一个向量后作为softmax函数的输入，然后输出得到对第j个状态变量估计值的索引表示为
其中f(x)为sigmoid函数的截断和放缩，可以表示为
其中x的定义域为0≤x≤1，参数p为放缩因子；表示对每一个g对应的取使最大的g；估计值可以表示为令表示对第j个状态变量估计值的可靠性指标，并设定观测次数的阈值和可靠性阈值；可以表示为
其中表示经过第l时隙的传输后，获得对第j个状态的观测次数，c为观测次数的阈值，表示第j个状态变量的第qj个观测值所对应的信道增益幅值；∈为可靠性阈值，则表示对第j个状态的观测是可靠的。 - 根据权利要求4所述的一种与源址无关的分布式状态监测方法，其特征在于，步骤S4具体包括：令则第k个传感器在第l+1个时隙内的激活概率可以表示为其中Ok表示第k个传感器能够观测到的状态变量索引集合，即表示Ok是{1,2,...,No}的子集；表示第j个状态变量的可靠性指标对第k个传感器在第l+1个时隙内激活概率的贡献权重整个待观测区域被分为Na个区域，第na个区域内传感器能观测到的状态索引集合为以一个区域为单位进行状态变量估计，如果一个区域内对所有可观测状态变量的估计都是可靠的，那么可以认为对这个区域的状态变量估计已完成；融合中心只需指示那些分布在其他区域内的传感器在下一个时隙内激活；在下一个时隙的传输中，希望尽可能少的区域激活就能使得所有待观测的状态变量被覆盖；将这个问题建模为一个集合覆盖的最佳化问题，如下所示：
矩阵表示不同区域内能观测到的状态变量；为矩阵B的元素，表示第na个区域内的传感器是否能获取对第j个状态变量的观测；表示选择第na个区域的代价，的值被设置为1；表示是否选择第na个区域；表示至少一个激活区域内的传感器获取对第j个状态变量的观测；采用贪婪算法求解这个集合覆盖的最佳化问题，令表示一个集合，该集合内的元素为在第l+1个时隙内需要继续观测的状态变量的索引；令表示一个集合，该集合内的元素为含有经过第l个时隙观测后不可靠状态变量估计区域的索引；在集合内，每次选取第na个区域,其中na可以表示为表示第na个区域内传感器能观测到的状态索引集合；然后将索引na从集合中移除并加入集合其中集合表示在第l+1个时隙内需要激活区域的索引集合；重复上述选择过程直到其中表示所有的na属于∪·表示对所有的取并集；最终将包含在第l+1个时隙内需要激活区域的索引；在一个传输时隙内，激活传感器的数量将被控制；分布在第个区域内的第个传感器的激活概率为表示第na个区域内的传感器索引集合，其中p0表示在第一个时隙内传感器的激活概率，具体数值要根据场景的情况设定；表示第l+1个时隙内第k个传感器的激活概率，根据得到；除了第na个区域之外的其他区域内的传感器在下一个传输时隙内保持休眠。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211201368.5 | 2022-09-29 | ||
| CN202211201368.5A CN115550994A (zh) | 2022-09-29 | 2022-09-29 | 一种与源址无关的分布式状态监测方法 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2024066284A1 true WO2024066284A1 (zh) | 2024-04-04 |
Family
ID=84732220
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2023/086658 WO2024066284A1 (zh) | 2022-09-29 | 2023-04-06 | 一种与源址无关的分布式状态监测方法 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115550994A (zh) |
| WO (1) | WO2024066284A1 (zh) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130218499A1 (en) * | 2010-07-27 | 2013-08-22 | Thales | Method for Optimally Determining the Characteristics and Arrangement of a Set of Sensors for Monitoring an Area |
| CN111835464A (zh) * | 2020-07-14 | 2020-10-27 | 浙江大学 | 一种与源址无关的大规模合作接入方法 |
| CN113194548A (zh) * | 2021-03-30 | 2021-07-30 | 浙江大学 | 一种智能超表面辅助的大规模无源随机接入方法 |
| CN114285523A (zh) * | 2021-12-24 | 2022-04-05 | 西安交通大学 | 面向多业务需求的大规模mtc免授权多用户检测方法及系统 |
| WO2022114561A1 (ko) * | 2020-11-27 | 2022-06-02 | 서울대학교산학협력단 | 심층 신경망을 이용한 활성 단말 검출과 채널 추정 방법 및 장치 |
-
2022
- 2022-09-29 CN CN202211201368.5A patent/CN115550994A/zh active Pending
-
2023
- 2023-04-06 WO PCT/CN2023/086658 patent/WO2024066284A1/zh unknown
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130218499A1 (en) * | 2010-07-27 | 2013-08-22 | Thales | Method for Optimally Determining the Characteristics and Arrangement of a Set of Sensors for Monitoring an Area |
| CN111835464A (zh) * | 2020-07-14 | 2020-10-27 | 浙江大学 | 一种与源址无关的大规模合作接入方法 |
| WO2022114561A1 (ko) * | 2020-11-27 | 2022-06-02 | 서울대학교산학협력단 | 심층 신경망을 이용한 활성 단말 검출과 채널 추정 방법 및 장치 |
| CN113194548A (zh) * | 2021-03-30 | 2021-07-30 | 浙江大学 | 一种智能超表面辅助的大规模无源随机接入方法 |
| CN114285523A (zh) * | 2021-12-24 | 2022-04-05 | 西安交通大学 | 面向多业务需求的大规模mtc免授权多用户检测方法及系统 |
Non-Patent Citations (1)
| Title |
|---|
| TAO QIN, CAIJUN ZHONG, ZHAOYANG ZHANG: "Ambient backscatter communications technology for batteryless IoT", CHINESE JOURNAL ON INTERNET OF THINGS, vol. 3, no. 2, 30 June 2019 (2019-06-30), pages 28 - 34, XP055839378, DOI: 10.11959/j.issn.2096−3750.2019.00107 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115550994A (zh) | 2022-12-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9628114B2 (en) | Length-compatible extended polar codes | |
| CN101621353B (zh) | 一种随机线性网络编码的方法、装置和系统 | |
| CN102651229B (zh) | 半导体装置和数据处理方法 | |
| CN111835464B (zh) | 一种与源址无关的大规模合作接入方法 | |
| CN104081701B (zh) | 光网络系统的通信方法、系统及装置 | |
| CN111447160B (zh) | 基于神经网络的大规模mimo无源随机接入的信号检测方法 | |
| CN105577329A (zh) | 基于空间调制的物理层安全传输方法 | |
| CN108494527A (zh) | 一种基于LoRa的数据发送和接收方法 | |
| KR20190117186A (ko) | 학습 기반 채널코드 디코딩 방법 및 장치 | |
| CN114499743A (zh) | 通信数据的处理方法、装置、设备及存储介质 | |
| WO2015180662A1 (zh) | 并行信道的编码调制方法和装置 | |
| WO2024066284A1 (zh) | 一种与源址无关的分布式状态监测方法 | |
| CN112996099B (zh) | 抗时隙删除的极化时隙aloha的发送与接收方法 | |
| CN106537959A (zh) | 电信网络中编码和解码帧的方法 | |
| CN117240409B (zh) | 一种用于智能手机与智能穿戴设备的数据处理方法 | |
| CN101754343B (zh) | 一种信道传输质量的反馈方法、系统和设备 | |
| KR20180117042A (ko) | 희소 코딩을 이용한 정보 전송 방법 및 장치 | |
| Wang et al. | Massive machine type communications: multiple access schemes | |
| WO2017185213A1 (zh) | 编码方法和编码装置 | |
| WO2022257718A1 (zh) | 一种极化码编码方法、译码方法及装置 | |
| WO2014032578A1 (zh) | 一种确定软比特信息的方法及装置 | |
| WO2018146554A1 (en) | Method and device for distributed error detection with polar codes | |
| CN115426077A (zh) | 译码路径确定方法、装置、计算机设备和存储介质 | |
| US10797726B2 (en) | Network data prediction method, network data processing device and network data processing method | |
| KR20100111628A (ko) | 그룹 변조 방법 및 이를 이용한 송신 장치 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 23869540 Country of ref document: EP Kind code of ref document: A1 |