CN107547456A - A kind of method for reducing link noise - Google Patents
A kind of method for reducing link noise Download PDFInfo
- Publication number
- CN107547456A CN107547456A CN201710504833.5A CN201710504833A CN107547456A CN 107547456 A CN107547456 A CN 107547456A CN 201710504833 A CN201710504833 A CN 201710504833A CN 107547456 A CN107547456 A CN 107547456A
- Authority
- CN
- China
- Prior art keywords
- components
- component
- local
- noise
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
A kind of method for reducing link noise, belong to communication technical field, solve influence of the link noise to method performance in sensor network, sensor node is in the form of vectors transmitting local local estimate to neighbor node, reduce the number of components to be transferred to neighbor node partial estimation value, then go to supplement the component not transmitted with the local estimation of aggregators, so just reduce link noise from source, if transmit one-component, only this one-component has noise, other components are to belong to local component, in the absence of link noise, then, consider to weaken the noise entrained by this one-component, so by the component with link noise and do not have noisy component to carry out an average calculating operation, meeting is weak much again after the influence of noise averagely, algorithm performance has obtained further raising.The present invention solves the problems, such as link noise well.
Description
Technical Field
The invention belongs to the technical field of communication, and particularly relates to a method for reducing link noise.
Background
At present, most of algorithms under the studied distributed diffusion cooperation strategy assume that links between nodes are ideal, namely, no link noise exists. However, in the real situation, the influence of link noise on the network is very large, which affects the self-estimated value transmitted to the next node, and the accuracy of the network for estimating the unknown parameter is greatly reduced by transmitting the affected estimated value to the wireless sensor network. In addition, link noise is likely to cause link failure, an incremental cooperation strategy which needs to form a ring in a distributed cooperation strategy is not suitable for the breakable network, and a diffusion cooperation strategy can be well compatible due to the network robustness.
1. Influence of link noise
1. DLMS algorithm under link noise condition
The communication link between sensor nodes is typically a noisy link. A typical example is time division multiple access, where a channel may be occupied and unavailable, even if marginally available, at a given time slot of a user, such a communication channel is noisy. The topological structure of the sensor network is always random, communication connection between nodes can be disconnected due to faults under the influence of link noise, the topological structure of the network can be recombined, and the research of the DLMS method under the distributed diffusion strategy in a noise link is of practical significance for improving the steady-state performance of the adaptive network.
The observation signals adopt a linear mathematical model: d is a radical of k (i)＝u k,i w 0 +v k (i) Wherein v is k (i) Is a variance ofAnd are additive white gaussian noise that are independent of each other in both time and space.
In order to obtain unknown parameters w of N node pairs in the network 0 Wherein w is the local estimate of 0 The vector is an M multiplied by 1 vector, and the mean square error of the vector is taken as an objective function (cost function) by utilizing a minimum mean square error criterion:
measurement data of all sensor nodes in the network d k,j ,u k,j Can be represented by the following matrix:the objective function (cost function) can be rewritten as:
DLMS algorithm in ideal link case: whereinIs the estimate of the unknown vector at time i-1 by node k,is a fused estimate at node k, μ k > 0 (constant) is the convergence step, non-negative c kl K =1,2, …, N is a local fusion parameter and satisfies
The node k receives the estimation value of target parameters of each node transmitted from the neighbor node (excluding the node k) in the self-adapting and fusion process under the condition that link noise exists between the node k and the node lBut some of these estimates have received M x 1 order additive noise vectorsThe DLMS algorithm in the case of a noisy link is shown in equation (1):
whereinAdditive link noise is a generalized stationary random process with a mean value of zero.
Unfolding formula (1) as follows:
order toEquation (2) can be rewritten as:
whereinRepresents and node kPhase (C)
Equivalent noise for all links connected. Therefore, the fusion process of the DLMS algorithm under the condition of considering noise is as follows:
the self-adaptive updating process of the self-estimated value of the node k is as follows:
2. DLMS method simulation under influence of noise link
Simulation conditions are as follows: and carrying out simulation analysis on the DLMS algorithm under the condition of existence of link noise by utilizing Matlab simulation software.
(1) The convergence step sizes are all set to μ k ＝0.05,k＝1,2,…,N；
(2)The additive noise is white gaussian noise and the signal-to-noise ratio SNR =10dB;
(3)Q k ＝10 -3 I M ；
(4) Taking global Mean Square Deviation (MSD) as a method performance index, namely MSD = E | | | ψ j -w (0) || 2 N, wherein ψ j And x 0 And respectively representing the estimated value and the unknown parameter of each node.
From the simulation results of fig. 2, it can be seen that, under the condition of not considering link noise, the MSD value of the whole sensor network is stabilized around-43 dB, the algorithm performance is good, and after considering link noise, the performance of the DLMS method is rapidly deteriorated, the MSD is stabilized around-28 dB, the MSD is directly deteriorated by 15dB, and the influence on the accuracy of completing unknown parameter estimation of the whole wireless sensor network is huge. This problem of link noise must be addressed in sensor networks to minimize its impact on the performance of the method.
In a wireless sensor network, all nodes cooperatively work by taking estimation of a common unknown parameter as a target, the nodes transmit data vectors acquired by the nodes to surrounding nodes according to random topologies under different strategies, and noise on a transmission link additively interferes the vectors. The link noise is mainly added into an estimated value signal in the transmission process, and due to the non-estimability of the link noise, the statistical characteristics and the size of the noise cannot be judged, and the noise cannot be removed from the signal.
Disclosure of Invention
The invention provides a method for reducing link noise in order to solve the influence of the link noise on the performance of a method in a sensor network.
The invention adopts the following technical scheme:
a method for reducing link noise starts from the component of the vector transmitted between sensor nodes, since the noise can not be directly removed, the number of the component of the estimation vector with noise transmitted from the neighbor nodes is reduced, the local component without link noise is used to replace the component without transmission, if only one noise component is transmitted, the rest is replaced by the local estimation vector component, the link noise does not exist in the local estimation, and thus the link noise is reduced from the source.
Setting parameters: mu.s k ＝0.05,k＝1,2,…,N；
Initialization: c. C l,k (0)＝1/|N k |，ψ k (0)＝0；
Further, a method of reducing link noise, comprising the steps of:
in a first step, a component selection matrix A is obtained by means of a pseudo-random number generator,
selecting a pseudo-random number generator to generate a sequence generation matrix M, wherein at the moment i, the local estimation value of the target parameter by the neighbor node l is an Mx 1 vectorNow onlyTransmitting L components thereof, and using local estimation values for the remaining M-L componentsFor M components, assume that the required transmission components are L (0)<L&M), replacing the rest M-L local estimation values with component values in the local estimation values, wherein the formula (3) is as follows:
wherein: c. C n，k Represents a fusion parameter of the neighbor node n and the local node k, andN k representing a set of neighbor nodes participating in a node k fusion operation, N k ＝{1,2,3,…,N}，An estimated value of a local node k at the time i is represented;m component representing the n node at time i, M =1,2 … M; n =1,2 … N;
second, select local estimate I M The component in A is filled in the local estimation value transmitted by the neighbor node, as shown in formula (4):
wherein:is a link noise vector;selecting a matrix;an estimate value representing a neighbor node l;
and step three, performing a fusion process and a self-adaptive updating process, wherein the fusion process is shown as a formula (5):
the component of vector A that is 0 is replaced by the corresponding component of the local estimate, i.e. the local estimate is replaced by the component of vector A that is 0Instead of the formerThen the fusion process is shown in equation (6):
wherein: phi is a k (i-1) Represents the fused estimate at time i-1, c nk Represents the fusion parameter of node N and local node k, N =1,2,3 … N k And is made of
The adaptive update procedure is shown in equation (7):
the obtained fusion estimated value phi k (i-1) Obtaining a local estimate Ψ by an adaptive update process k (i) The local estimation value is sent to the neighbor node again for calculating the fusion estimation value of the neighbor node;
fourthly, weakening the noise carried by the components, averaging the components with link noise and the components without noise, and making neighboring nodesSum of local estimate components at node k for point lAs shown in formula (8):formula (8), wherein I 1×M The vector is used for adding all components participating in fusion, and after averaging, the average value of the components isDefining local estimates of fusion stage nodes iThen the fusion process with averaging is shown as equation (9):
in a second step I of local estimated values is selected M The method of selecting the components in a comprises selecting the components to be transmitted randomly or in a sequential manner.
Randomly selecting a component to be transmitted, comprising the steps of: the nodes use the pseudo-random number generator to generate random sequence to select the components to be transmitted, after the neighbor nodes I select the components to be transmitted, a seed file is generated, and the coordinate information of the transmitted components, namely the coordinate information of the transmitted components is recordedAnd defining a component selection matrix, wherein 1 represents a transmission component, 0 represents that the component is not transmitted, and the node k fills the non-transmission component in the neighbor node with the corresponding component in the local estimation value according to the coordinate information.
The method for selecting the components to be transmitted in a sequential manner comprises the following steps: and selecting and utilizing components in the sequence of each neighbor node according to the sequence containing all component coordinates in one period, namely two transmissions, transmitting a seed containing component coordinate information when transmitting the local estimation value, and transmitting all component information of the neighbor nodes to the node k for fusion after multi-period transmission. The whole process is as follows: assuming that the total number of components to be transmitted by the neighboring node is 4, the pre-customized sequence may be (1,0,1,0), (0,1,0,1), so that after one period (two transmissions), the local node k completely acquires the component information of the neighboring node.
The invention has the following beneficial effects:
the invention reduces the link noise from the source, the improved method well solves the problem of the link noise, in order to weaken the noise carried by one component, the component with the link noise and the component without the noise are subjected to an average operation, the influence of the noise is much weaker after averaging, and the method performance is further improved.
Drawings
FIG. 1 is a diagram of a network topology;
FIG. 2 is a DLMS algorithm simulation analysis diagram under the influence of a noise link;
FIG. 3 is a transmission process of a local estimate component;
FIG. 4 is a simulation analysis diagram of the present invention;
FIG. 5 is a simulation analysis diagram of an improved algorithm for component averaging.
Detailed Description
Two ways of selecting the component of the local estimation value to be transmitted by the neighbor node are available, one is random selection, the node selects the component to be transmitted by using a pseudo-random number generator to generate a random sequence, in this way, after the neighbor node I selects the component to be transmitted, a seed file is generated, and the coordinate information (namely, the coordinate information of the transmitted component) is recordedDefined as a component selection matrix, where 1 represents a transmitted component and 0 represents no transmission of that component), node k is based on the co-ordinatesThe beacon information fills in the untransmitted components in the neighboring nodes with the corresponding components in the local estimate.
Taking the number of nodes N =8 and the total number of components M =4 as an example, the number of components L =2 selected by the node 1,2 in a random manner is shown in the formula (10) and the formula (11), and after 8 time instants, all component information of the node 1,2 is transmitted to the node k.
Another method is to select the components to be transmitted in a sequential manner, i.e. select and utilize some components of each neighbor node according to a predetermined sequence, this manner will also transmit a seed containing component coordinate information when transmitting the local estimation value, the same condition of the former method is analyzed as shown in formula (12) and formula (13), and similarly, after transmission at a certain time, all component information of the neighbor nodes can be transmitted to the node k for fusion.
The method of the invention is subjected to simulation analysis, and the simulation conditions are as follows: matlab simulation software is used for simulating the DLMS method according to the conditions of whether link noise exists and whether components are averaged or not. The detailed parameters of the method are as follows:
(1) The convergence step sizes are all set to μ k ＝0.05,k＝1,2,…,N；
(2)The additive noise is white gaussian noise and the signal-to-noise ratio SNR =10dB;
(3)Q k ＝10 -3 I M ；
(4) Taking global Mean Square Deviation (MSD) as a method performance index, namely MSD = E | | | ψ j -w (0) || 2 N, wherein ψ j And x 0 And respectively representing the estimated value and the unknown parameter of each node.
As shown in fig. 4, a simulation analysis of the performance of the present invention and the conventional DLMS algorithm is given in consideration of link noise. The method has good performance improvement, the performance of the traditional DLMS algorithm is sharply deteriorated when the noise is transmitted completely, the MSD value is reduced to-25 dB, and even the performance is lower than that of single node fusion; the invention adopts the local estimation value component to replace part of noise component, the value of MSD is stabilized at-33 dB, the performance value of single node fusion is achieved, and compared with-38 dB of noise-free part transmission, certain noise influence still exists. Under the condition of no noise, the performance stability value of partial transmission is higher than that of full transmission by about 2dB, which shows that the partial transmission still has certain defects on the whole system performance.
As shown in fig. 5, a simulation result of averaging a part of the noise component and a part of the local noise-free component and participating in the fusion process using the average value is shown. The method after averaging has a certain improvement on convergence speed, reaches a stable value after 60 iterations, overcomes the noise influence, further improves the noise influence under the condition of partial full transmission, and reaches the vicinity of-40 dB of the DLMS method without considering link noise. The method performance is greatly improved.
Claims (5)
1. A method of reducing link noise, characterized by: starting from the components of vectors transmitted between sensor nodes, the number of the components of estimation vectors with noise transmitted from neighbor nodes is reduced, local components without link noise are adopted to replace components without transmission, only one noise component is transmitted, and the rest components are replaced by local estimation vector components, so that the link noise is reduced from the source.
2. A method for reducing link noise according to claim 1, wherein: the method comprises the following steps:
in a first step, a component selection matrix A is obtained by means of a pseudo-random number generator,
selecting a pseudo-random number generator to generate a sequence generation matrix M, wherein at the moment i, the local estimation value of the target parameter by the neighbor node l is an Mx 1 vectorOnly L of these components are now transmitted, the remaining M-L components using local estimatesFor M components, the number of required transmission components is L, and the remaining M-L components are replaced by component values in the local estimation value, as shown in the following formula:
wherein:
c n,k represents the fusion parameters of neighbor node N and local node k, and N =1,2,3 … N,N k representing a set of neighbor nodes participating in a node k fusion operation, N k ＝{1,2,3,…,N}，An estimated value representing a local node k at time i;m component representing the n node at time i, M =1,2 … M; n =1,2 … N;
second, selecting local estimated value I M The component in A is filled in the local estimation value transmitted from the neighbor node, as shown in the following formula:
wherein:
is a link noise vector;selecting a matrix;an estimate value representing a neighbor node l;
and thirdly, performing a fusion process and a self-adaptive updating process, wherein the fusion process is shown as the following formula:
the component of vector A having a value of 0 is replaced by the corresponding component of the local estimate, i.e. the local estimate is replaced by the component of vector A having a value of 0Instead of the formerThe fusion process is:
wherein: phi is a k (i-1) Represents the fused estimate at time i-1, c nk Represents the fusion parameter of node N and local node k, N =1,2,3 … N k And is and
the adaptive update procedure is shown as follows:
the obtained fusion estimated value phi k (i-1) Obtaining local estimates by adaptive updatingThe local estimation value is sent to a neighbor node again for calculating a fusion estimation value of the neighbor node;
fourthly, weakening the noise carried by the components, carrying out an average operation on the components with link noise and the components without the noise, and carrying out the sum of local estimation value components of the neighbor node l at the node kComprises the following steps:wherein I 1×M The vector function is to add all the components participating in the fusion, and after averaging, the average value of the components isDefining local estimates of fusion stage nodes iThe fusion process with averaging is then:
3. a method of reducing link noise according to claim 2The method is characterized in that: in a second step I of local estimated values is selected M The method of selecting the components in a includes selecting the components to be transmitted randomly or in a sequential manner.
4. A method for reducing link noise according to claim 3, wherein: the random selection of the components to be transmitted comprises the steps of: the nodes use the pseudo-random number generator to generate random sequence to select the components to be transmitted, after the neighbor nodes I select the components to be transmitted, a seed file is generated, and the coordinate information of the transmitted components, namely the coordinate information of the transmitted components is recordedAnd defining a component selection matrix, wherein 1 represents a transmission component, 0 represents that the component is not transmitted, and the node k fills the non-transmission component in the neighbor node with the corresponding component in the local estimation value according to the coordinate information.
5. A method for reducing link noise according to claim 3, wherein: the sequential manner selects the components to be transmitted, comprising the steps of: selecting and utilizing components in the sequence of each neighbor node according to the sequence containing all component coordinates in one period, namely two transmissions, transmitting a seed containing component coordinate information when transmitting a local estimation value, and transmitting all component information of the neighbor nodes to a node k for fusion after multi-period transmission.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710504833.5A CN107547456B (en) | 2017-06-27 | 2017-06-27 | Method for reducing link noise |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710504833.5A CN107547456B (en) | 2017-06-27 | 2017-06-27 | Method for reducing link noise |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107547456A true CN107547456A (en) | 2018-01-05 |
| CN107547456B CN107547456B (en) | 2020-03-31 |
Family
ID=60970309
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710504833.5A Active CN107547456B (en) | 2017-06-27 | 2017-06-27 | Method for reducing link noise |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107547456B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101933289A (en) * | 2008-01-31 | 2010-12-29 | 微软公司 | The management of wireless network |
| US20120135691A1 (en) * | 2010-11-29 | 2012-05-31 | King Fahd University Of Petroleum And Minerals | Noise-constrained diffusion least mean square method for estimation in adaptive networks |
| CN103313386A (en) * | 2013-06-18 | 2013-09-18 | 浙江大学 | Method for tracking targets of wireless sensor network on basis of information consistency weight optimization |
-
2017
- 2017-06-27 CN CN201710504833.5A patent/CN107547456B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101933289A (en) * | 2008-01-31 | 2010-12-29 | 微软公司 | The management of wireless network |
| US20120135691A1 (en) * | 2010-11-29 | 2012-05-31 | King Fahd University Of Petroleum And Minerals | Noise-constrained diffusion least mean square method for estimation in adaptive networks |
| CN103313386A (en) * | 2013-06-18 | 2013-09-18 | 浙江大学 | Method for tracking targets of wireless sensor network on basis of information consistency weight optimization |
Non-Patent Citations (3)
| Title |
|---|
| 卢光跃等: "基于信誉机制的分布式扩散最小均方算法", 《电子与信息学报》 * |
| 郭龙江等: "传感器网络感知数据自适应去噪方法", 《计算机工程与应用》 * |
| 魏巍等: "恶意攻击下改进的分布式扩散最小均方算法", 《西安邮电大学学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107547456B (en) | 2020-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8743759B2 (en) | Detection of access point location errors in enterprise localization systems | |
| CN107395440B (en) | Internet topology detection node optimization deployment method based on complex network | |
| Kim et al. | Cognitive radio spectrum prediction using dictionary learning | |
| WO2021036414A1 (en) | Co-channel interference prediction method for satellite-to-ground downlink under low earth orbit satellite constellation | |
| CN114666204B (en) | Fault root cause positioning method and system based on causal reinforcement learning | |
| Mukhopadhyay et al. | Model based error correction for wireless sensor networks | |
| CN109104307B (en) | Key node sensing method of dynamic data link network | |
| Campagnaro et al. | A simulation framework for smart adaptive long and short-range acoustic networks | |
| CN107547456B (en) | Method for reducing link noise | |
| CN111194048B (en) | EM-based 1-bit parameter estimation method | |
| Yang et al. | ESPN: Efficient server placement in probabilistic networks with budget constraint | |
| CN112183425A (en) | Channel parameter estimation method and device | |
| CN112865891A (en) | PUCCH signal detection method, apparatus, communication device and storage medium | |
| Vasconcelos et al. | Remote estimation games over shared networks | |
| Alonso-Román et al. | Adaptive consensus-based distributed Kalman filter for WSNs with random link failures | |
| CN108594169B (en) | Multi-robot distributed cooperative positioning method suitable for time-varying communication topology | |
| Pottier et al. | Q-learning based adaptive channel selection for underwater sensor networks | |
| Diniz et al. | Data censoring with set-membership algorithms | |
| CN111683377B (en) | Real-time reliable relay deployment method for power distribution network | |
| Li et al. | A distributed variable tap-length algorithm within diffusion adaptive networks | |
| Chen et al. | Performance analysis of diffusion LMS in multitask networks | |
| CN113078885A (en) | Distributed adaptive estimation method for resisting pulse interference | |
| KR102207144B1 (en) | Methods and apparatuses for generating real-environment noise using statistical approach | |
| CN108400948B (en) | Environment self-adaptive perception wireless communication channel estimation and signal reconstruction method | |
| Listova et al. | The communications channels models in wireless sensor networks, based on the structural-energetic interaction between signals and interferences |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 03060606 No. 280 Wenyuan Street, Yuci District, Jinzhong City, Shanxi Province Applicant after: China Railway Third Engineering Co., Ltd. Applicant after: The Electricication Engineering Co.,Ltd.Of The Fhird Engineering Group Co.,Ltd.Of China Railway Address before: 030001 Yingze street, Shanxi, No. 269, No. Applicant before: China Railway Third Engineering Co., Ltd. Applicant before: The Electricication Engineering Co.,Ltd.Of The Fhird Engineering Group Co.,Ltd.Of China Railway |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |