CN114501331B

CN114501331B - Self-adaptive physiological monitoring and intelligent scheduling positioning system and method

Info

Publication number: CN114501331B
Application number: CN202111650175.3A
Authority: CN
Inventors: 卢旭; 陈可洲; 刘军
Original assignee: Guangdong Polytechnic Normal University
Current assignee: Guangdong Polytechnic Normal University
Priority date: 2021-12-30
Filing date: 2021-12-30
Publication date: 2023-06-23
Anticipated expiration: 2041-12-30
Also published as: CN114501331A

Abstract

The invention relates to a self-adaptive physiological monitoring and intelligent scheduling positioning system and a self-adaptive physiological monitoring and intelligent scheduling positioning method, wherein the system comprises a physiological data monitoring module and a rescue module; the physiological data monitoring module collects physiological data of a human body by utilizing a somatosensory network, divides the priority level of the data by an RM-MAC (remote management-media access control) improvement protocol and performs self-adaptive transmission by an RS (Reed-Solomon) algorithm; the rescue module is used for assisting medical staff in efficiently positioning the patient by utilizing a multi-source heterogeneous indoor sensing positioning and sensing node advanced scheduling technology when the physiological data of the user is abnormal, so that the patient is timely rescued, and the energy consumption of the indoor heterogeneous sensing positioning network is saved. The invention applies the deep learning target track prediction technology to the indoor rescue module, carries out high-performance scheduling on the position tracking nodes in advance through track prediction, ensures the perception tracking precision, reduces the network energy consumption, and can carry out effective and rapid position rescue to avoid risks when the system monitors abnormal physiological data which possibly endanger the life of a patient.

Description

Self-adaptive physiological monitoring and intelligent scheduling positioning system and method

Technical Field

The invention relates to the technical field of physiological monitoring and pedestrian track prediction, in particular to a self-adaptive physiological monitoring and intelligent scheduling positioning system and method.

Background

With the development of wireless wearable biosensors, more heterogeneous wireless wearable biosensors are used in combination to identify and monitor complex physiological states of the human body, which are called low-power and low-cost Wireless Body Area Networks (WBANs). WBSNs have a number of fields of application such as healthcare, military, recreational and sports. Currently, the application field of WBANs is mainly health monitoring. With the increasing demand for high performance WBANs, the advent of IEEE 802.15.6, a communication standard specifically designed for WBANs, resulted. Scholars have designed a number of algorithms and communication protocols with advanced capabilities based on the IEEE 802.15.6 standard to further improve the Network Lifetime (NL) and quality of service (QoS) of WBAN. While most students treat data between normal and highly abnormal as a single category, such data is large in size and conventional routing protocols tend to result in the loss of more important, valuable data. In addition, WBAN is rapidly developing in medical monitoring and diagnostics, but most scholars do not consider taking corresponding emergency rescue measures when an emergency event is detected. However, corresponding emergency rescue measures are necessary. Therefore, an Indoor Rescue Module (IRM) is designed by innovatively utilizing a Wireless Sensor Network (WSN) indoor positioning technology and a deep learning pedestrian track prediction technology. However, the indoor positioning scheme based on the WSN also has the problems of limited energy, difficult replenishment and the like, and the most common existing method for prolonging the service life of the WSN network is sleep scheduling, and partial redundant nodes are put into a dormant state under the condition of ensuring the coverage rate of a monitored area, and then the node state is updated through a wake-up algorithm. However, this method may deteriorate the accuracy of positioning tracking. It is very difficult to perform rescue work in case of insufficient positioning accuracy, and even the best rescue time of the patient may be missed.

Therefore, how to provide an adaptive physiological monitoring and intelligent scheduling positioning system APMIPS with high efficiency, high intelligence and capability of providing more timely and higher life support for patients in emergency and providing more accurate rescue information for medical staff is a problem to be solved by the technicians in the field.

Disclosure of Invention

In order to solve the technical problems in the prior art, the invention provides a self-adaptive physiological monitoring and intelligent scheduling positioning system and a method, which apply a deep learning target track prediction technology to an indoor rescue module, schedule position tracking nodes in advance through track prediction with high performance, ensure the perceived tracking precision, reduce network energy consumption, and when the system monitors abnormal physiological data which possibly endanger the life of a patient, the rescue module can implement effective and rapid position rescue to avoid risks.

The system is realized by adopting the following technical scheme: the self-adaptive physiological monitoring and intelligent scheduling positioning system comprises a physiological data monitoring module and a rescue module, wherein:

the physiological data monitoring module collects physiological data of a human body by utilizing a somatosensory network, divides the priority level of the data by an RM-MAC (remote management-media access control) improvement protocol, and finally performs self-adaptive transmission by an RS (Reed-Solomon) algorithm;

the rescue module is used for assisting medical staff in efficiently positioning the patient by utilizing a multi-source heterogeneous indoor sensing positioning and sensing node advanced scheduling technology when the physiological data of the user is abnormal, so that the patient is timely rescued, and the energy consumption of the indoor heterogeneous sensing positioning network is saved.

The method is realized by adopting the following technical scheme: a self-adaptive physiological monitoring and intelligent scheduling positioning method comprises the following steps:

s1, acquiring physiological data of a human body by using a somatosensory network, entering a rescue module if the data type of the acquired data is 11, enabling the somatosensory network to enter a continuous perception state, and distributing an independent communication channel for a user; otherwise, entering a step S2;

s2, if the data type of the collected data is 10, the collected data is directly transmitted to the sink node; otherwise, entering a step S3;

s3, if the data type of the acquired data is 01, primarily judging the priority level of the monitored data, sending a transmission request to the sink node by the somatosensory network, acquiring information of all adjacent nodes, calculating the RS value of the adjacent nodes by the sink node by adopting an RS algorithm, and then selecting the node with the maximum RS value as a forwarding relay node of the sink node, wherein the relay node transmits the data to the sink node;

s4, if the sink node checks that the self residual energy is lower than a set threshold value, the work is ended, and otherwise, the next sensing period is started.

Compared with the prior art, the invention has the following advantages and beneficial effects:

1. according to the invention, the emergency rescue function based on indoor positioning is incorporated into the traditional physiological monitoring system by utilizing the RM-MAC improved protocol, and the RS algorithm is utilized to select proper relay forwarding according to the importance of physiological data, so that the priority transmission of more important monitoring data is ensured.

2. The invention applies the deep learning target track prediction technology to the indoor rescue module, carries out high-performance scheduling on the position tracking nodes in advance through track prediction, reduces network energy consumption while ensuring the perceived tracking precision, and can carry out effective and rapid position rescue to avoid risks when the system monitors abnormal physiological data which possibly endanger the life of a patient.

Drawings

FIG. 1 is a block diagram of a system of the present invention;

FIG. 2 is a diagram of the RM-MAC improvement protocol framework of the present invention;

FIG. 3 is a schematic illustration of a node prescheduling based on trajectory prediction in accordance with the present invention;

FIG. 4 is a schematic diagram of a frame of an IRM module;

FIG. 5 (a) is a comparative schematic diagram of network lifetime of a protocol algorithm;

fig. 5 (b) is a schematic diagram showing comparison of data transmission delays of a protocol algorithm;

FIG. 5 (c) is a diagram illustrating a comparison of network throughput for a protocol algorithm;

FIG. 6 (a) is a schematic diagram showing a comparison of network lifetime performance of a priority-based RS algorithm;

fig. 6 (b) is a schematic diagram showing comparison of data transmission delay performance of the RS algorithm based on priority;

FIG. 6 (c) is a schematic diagram showing a comparison of the throughput performance of the RS algorithm data network based on priority;

FIG. 7 (a) is a diagram showing the comparison of network lifetime of a protocol algorithm at different node densities;

FIG. 7 (b) is a schematic diagram showing the comparison of data transmission delays of a protocol algorithm at different node densities;

FIG. 7 (c) is a diagram showing the throughput versus the protocol algorithm at different node densities;

FIG. 8 (a) is a schematic diagram of a predictive scheduling scenario for a sensing node of an ETH dataset;

FIG. 8 (b) is a schematic diagram of a predictive scheduling scenario of a sensor node of the HOTEL dataset;

FIG. 8 (c) is a schematic diagram of a predictive scheduling scenario for a sensing node of a ZARA1 dataset;

FIG. 8 (d) is a schematic diagram of a predictive scheduling scenario for a sensing node of a ZARA2 dataset;

FIG. 8 (e) is a diagram of a predicted scheduling scenario for a sensing node of a UNIV dataset;

fig. 9 is a schematic diagram of a summary of advanced scheduling performance of a wireless sensor node;

fig. 10 is a flow chart of the method of the present invention.

Detailed Description

The present invention will be described in further detail with reference to examples and drawings, but embodiments of the present invention are not limited thereto.

Examples

As shown in fig. 1, the adaptive physiological monitoring and intelligent scheduling positioning system of the present embodiment includes a physiological data monitoring module and a rescue module, wherein:

the rescue module is used for assisting medical personnel in efficiently positioning the patient by utilizing a multi-source heterogeneous indoor sensing positioning and sensing node advanced scheduling technology when physiological data of the user are abnormal or even endanger life, so that the patient is timely rescued, and meanwhile, the energy consumption of the indoor heterogeneous sensing positioning network is saved.

Specifically, in this embodiment, the RM-MAC improvement protocol is a protocol that is improved on the basis of the RM-MAC protocol, and the indoor positioning rescue function is integrated into the physiological monitoring system by adding a rescue type data packet in the superframe structure of the MAC protocol;

the RS algorithm is a relay node selection algorithm specifically designed for the RM-MAC improvement protocol, and is used to adaptively select an appropriate data forwarding relay node according to the priority level of the data.

The advanced scheduling technology of the sensing node predicts the track of the pedestrian by utilizing the existing deep learning pedestrian track prediction technology, and then performs advanced scheduling on the sensing node by utilizing the combination of the predicted track and the adaptive scheduling radius ACR;

the self-adaptive scheduling radius ACR is the sum of the size of a confidence interval obtained by analyzing the errors of the predicted track and the real track through a probability analysis technology and the minimum distance required for realizing the coordinate positioning of the predicted track;

all the technologies and algorithms related to the physiological data monitoring module and the rescue module are used for information interaction and operation in physiological monitoring.

In this embodiment, the physiological data monitoring module is composed of WBANs composed of various wearable wireless biological sensors, RM-MAC improvement protocol and RS algorithm.

Research shows that the MAC protocol plays an important role in improving the reliability and energy efficiency of network information transmission. The MAC frame consists of a frame header, a frame body and a frame check sequence FCS. The MAC frame header consists of four fields, the first field being a control field, consisting of four 8-bit bytes. Modified-MAC (M-MAC) is a Modified version of the MAC protocol, with a data type field added to the MAC header, while the header length remains unchanged. The data types of the RM-MAC are expanded into four types, as shown in fig. 2, namely normal data ND, high normal data HD, critical data CD, and rescue data RD, and further refine the priority of the HD packet to reduce the loss rate of high-value data. The data with different priority levels can be transmitted in different transmission modes, RD data is transmitted in an independent channel, CD data is directly transmitted to the sink node, HD data is transmitted in a multi-hop mode, and ND data is not transmitted. Specifically, the expression and transmission method of the data are shown in table 1:

TABLE 1

Data type	Binary expression form	Transmission method
			Normal state	00	Not to send
Abnormality of	01	Using RS
			Severe severity of	10	Direct transmission
Rescue device	11	Independent channel

However, there is still a shortage of classified transmission of data, and HD packets that are more important in multi-hop transmission may be discarded during transmission because of the expiration date being exceeded. The labeling of the data priority is not clear in many researches, the importance of the data packets of the same type is different, and the transmission of the data with high importance (HD data with high abnormality degree) is ensured to be more beneficial to disease prevention and diagnosis. In addition, system lifetime and monitoring performance are two conflicting factors. Reducing energy consumption affects monitoring performance, while ensuring high monitoring performance requires more energy consumption. The RS algorithm can ensure the transmission of data with higher importance according to actual conditions, balance the service life and monitoring performance of the system and improve the intelligent degree of the physiological monitoring of the system. Specifically, the RS algorithm is defined as follows:

wherein E is _i Is the residual energy of the relay node, L _i Is the real-time load of the relay node, alpha is a constant factor, k is the priority of similar data, D _i Is the real-time transmission delay of the relay node.

In this embodiment, as shown in fig. 4, the rescue module is composed of a WSN indoor positioning and tracking module, a pedestrian track prediction network module, and an adaptive dispatching radius ACR calculation module. The WSN indoor positioning tracking module acquires a section of track of a target, inputs the track to the trained pedestrian track prediction network module, and performs advanced scheduling on peripheral nodes according to the predicted track through the adaptive scheduling radius ACR calculation module.

The WSN indoor positioning tracking technology has been developed quite mature, so the focus of rescue module research is not a tracking scheme, but a scheduling method of tracking nodes. In the rescue module, a sparse graph rolling network SGCN is used as a network framework for pedestrian track prediction. Unlike existing vision-based pedestrian trajectory prediction networks, the prediction network of the rescue module may take as input the trajectories acquired by the wireless sensing nodes. The method may be more global than vision-based methods.

During track prediction, the error increases with the prediction length. The prediction track output by the sparse graph convolution network SGCN is transmitted to an error analysis module to obtain the size of an error confidence interval where each prediction coordinate is located, so that the adaptive scheduling radius ACR calculation module can adaptively adjust the node pre-scheduling range according to the prediction step length. The prescheduling range is to activate all dormant nodes within the adaptive scheduling radius ACR with the prediction coordinates as the center of the circle, and the set of these nodes is called a prescheduling set PS. The actual coordinates of the pedestrians also correspond to a set of nodes to be scheduled, the set of which is called the actual scheduling set AS. To more clearly show the performance of ACR node based scheduling methods, two indexes are defined, one is scheduling accuracy SA and the other is radius ratio RR. The scheduling accuracy SA is the ratio between the number N of successful schedules and the total number TN of schedules, when

Such pre-scheduling is successful and the radius ratio is the ratio to the adaptive scheduling radius ACR. Specifically, the adaptive scheduling radius ACR calculation is defined as follows:

wherein set []Is the set of errors between the actual coordinates and the predicted coordinates, sort () is the ranking function,

() The confidence interval size solving function is used for calculating the difference between the upper limit and the lower limit of the confidence interval, and the significance level alpha=0.05; d (D) _min Is the shortest distance to successfully locate the predicted coordinates. IRM adopts classical three-point positioning algorithm, so D _min ＝max(dis(N[n ₁ ,n ₂ ,n ₃ ]) As shown in fig. 3). N in the above]Is the set of scheduling nodes needed to successfully locate the predicted coordinates, and the function dis () is the distance between the calculated predicted coordinates and the node.

The experiment of the invention is divided into two parts, namely a physiological monitoring data processing experiment and a wireless sensing node scheduling experiment, wherein the physiological monitoring data processing experiment is mainly used for comparing the performance of a communication protocol, and the wireless sensing node scheduling experiment is mainly used for evaluating the scheduling effect of the node.

1. Processing physiological monitoring data;

the overall performance of LBEE, WEQ, SIMPLE and RS in the most energy efficient mode was compared, as shown in fig. 5 (a), with the longest network lifetime for RS and LBEE algorithms and the shortest WEQ algorithm. This is because the WEQ algorithm tends to select nodes closer to the sink node to forward data, resulting in rapid energy exhaustion and death of those nodes. As shown in fig. 5 (b), the transmission delays of the RS, LBEE and WEQ algorithms are not very different, but the data transmission delay of the SIMPLE algorithm is much higher than the other three algorithms because the SIMPLE algorithm does not consider the forwarding delay of the node when selecting the next hop routing node. As shown in fig. 5 (c), the throughput of the RS algorithm is highest, and the throughput of the RS and LBEE algorithms is much higher than those of WEQ and SIMPLE algorithms. Among all the comparison algorithms, the RS algorithm has the highest overall performance.

Data transmission delay and throughput are important parameters of QoS, as shown in fig. 6, the RS performance affects the data response of different priorities (k). As shown in fig. 6 (a), when a lower transmission delay is provided for data with a larger k value, the energy of the network is intensively lost, resulting in a reduction of 28.33% in network lifetime. However, fig. 6 (b) and 6 (c) show that as network lifetime decreases, data transmission delay decreases by 30.77% and throughput increases by 14.94%. The data transmission delay and throughput increase rate are the highest phases, reaching 26.25% and 15.85%, respectively.

Network architecture compatibility is also an important criterion in view of network operation performance, and fig. 7 shows the performance of RS, LBEE, WEQ and SIMPLE algorithms at different node densities. As shown in fig. 7 (a), the RS and LBEE algorithms maintain the highest and most stable network lifetime despite the increased network node density, followed by the SIMPLE algorithm, while the WEQ algorithm has the shortest network lifetime. As shown in fig. 7 (b), at different node densities, the transmission delay of SIMPLE is highest, the data transmission delay variation of WEQ is relatively stable, and the data transmission delay of RS and LBEE algorithm is kept at a low level, and drops sharply with increasing node density. When x=30, the delay of the RS algorithm is lowest. As shown in fig. 7 (c), the throughput of WEQ and SIMPLE algorithms remains stable, while the throughput of RS and LBEE algorithms increases with increasing node density. Among them, the RS has the highest growth rate. Analysis of fig. 7 (a), (b) and (c) shows that the RS algorithm maintains stable network lifetime, reduces transmission delay by about 60%, and increases network throughput by 640% with increasing network node density. It is more flexible than the LBEE algorithm because the data packet can adaptively select a more appropriate transmission path. In addition, the RS algorithm is more suitable for being applied to a network with high node density, and the characteristic completely accords with the trend of network complexity and high performance of WBANs. Specifically, detailed performance comparisons of the physiological monitoring module versus the algorithm are shown in table 2:

TABLE 2

2. Scheduling experiments by the wireless sensing nodes;

to verify the effectiveness of the proposed method, two public pedestrian trajectory data sets were first used to train the SCCN network, ETH and UCY, which are the most widely used trajectory references. We neglect the actual environment when ETH and UCY are acquired and put these trajectories on an unobstructed planar space with wireless sensing nodes evenly distributed on these planes. As shown in fig. 8 (a), 8 (b), 8 (c), 8 (d) and 8 (e), the prediction scheduling conditions of the sensing nodes of five different data sets are shown, solid points uniformly distributed in the graph represent the sensing nodes, except for the range covered by a circle and the nodes for forwarding data being in an active state, other nodes are in a dormant state, and other nodes work normally. In addition, the input track for track prediction in the figure is also omitted, and the range covered by the small circle represents the scheduling range in an ideal state. The range covered by the small circle in the figure represents the minimum node coverage required for locating the real coordinates, and the range covered by the large circle represents the coverage of the pre-scheduled node. The error of track prediction is accumulated with the increase of the predicted track length, so that the range of advanced scheduling is also enlarged. The advanced scheduling range of the sensing node is related to the predicted quality of the target track, and the smaller the error of the predicted track is, the smaller the advanced scheduling range is, and the higher the energy efficiency of the network is. The pre-scheduling results of the five data sets show that the self-adaptive confidence scheduling radius can well realize the pre-scheduling of the nodes even if the track prediction accuracy is not high enough. In the figure, the track segments of ETH and ZARA1 have the best pre-scheduling effect, and the pre-scheduling effect of ZARA2 is slightly insufficient, but according to the coverage condition of the pre-scheduling nodes, the pre-scheduling can be judged to still realize the positioning of real coordinates. Furthermore, the scope of pre-scheduling is related to the density of the spatial sensing nodes being monitored. The SA and RR of the five data sets ETH, HOTEL, NUIV, ZARA1, ZARA2 are shown in fig. 9. The SA values of all five data sets exceeded 90% and were relatively smooth, but the RR values were not high, between 0.4 and 0.6, which clearly relates to the quality of the trajectory prediction. Therefore, in order to realize more accurate sensor node pre-scheduling, improving the accuracy and stability of track prediction is a key break.

Based on the same inventor conception, the self-adaptive physiological monitoring and intelligent scheduling positioning method of the embodiment comprises the following steps:

s3, if the data type of the acquired data is 01, primarily judging the priority level of the monitored data, wherein the priority level is arranged to be 11>10>01>00 from high to low, the somatosensory network sends a transmission request to the sink node, information of all adjacent nodes is obtained, the sink node calculates the RS value of the adjacent nodes by adopting an RS algorithm, then the node with the largest RS value is selected as a forwarding relay node, and the relay node transmits the data to the sink node;

s4, if the sink node checks that the self residual energy is lower than a set threshold value, namely 5% of total energy, the work is ended, and otherwise, the next sensing period is started.

The above examples are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above examples, and any other changes, modifications, substitutions, combinations, and simplifications that do not depart from the spirit and principle of the present invention should be made in the equivalent manner, and the embodiments are included in the protection scope of the present invention.

Claims

1. The self-adaptive physiological monitoring and intelligent scheduling positioning system is characterized by comprising a physiological data monitoring module and a rescue module, wherein:

the rescue module is used for assisting medical personnel in efficiently positioning the patient by utilizing a multi-source heterogeneous indoor sensing positioning and sensing node advanced scheduling technology when the physiological data of the user is abnormal, so that the patient is timely rescued, and the energy consumption of an indoor heterogeneous sensing positioning network is saved;

the advanced scheduling technology of the sensing node predicts the track of the pedestrian by using the deep learning pedestrian track prediction technology, and then performs advanced scheduling on the sensing node by combining the predicted track with the adaptive scheduling radius ACR;

all technologies and algorithms in the physiological data monitoring module and the rescue module are used for information interaction and operation in physiological monitoring;

the frame head of the RM-MAC improvement protocol is added with a data type field, and the field is expanded into four types, including normal data ND, high normal data HD, key data CD and rescue data RD; the method comprises the steps that normal data ND is not sent, high normal data HD is forwarded by a relay node through an RS algorithm, key data CD is directly sent to a sink node, and rescue data RD is sent through an independent channel;

the definition of the RS algorithm is as follows:

wherein E is _i Is the residual energy of the relay node, L _i Is the real-time load of the relay node, alpha is a constant factor, k is the priority of similar data, D _i Is the real-time transmission delay of the relay node;

the rescue module consists of a WSN indoor positioning tracking module, a pedestrian track prediction network module and a self-adaptive dispatching radius ACR calculation module; the WSN indoor positioning tracking module acquires a section of track of a target, inputs the track of a trained pedestrian into the prediction network module, and performs advanced scheduling on peripheral nodes according to the prediction track through the adaptive scheduling radius ACR calculation module;

the rescue module is used for taking the sparse graph rolling network SGCN as a network frame for pedestrian track prediction; the network frame of pedestrian track prediction takes the track acquired by the wireless sensing node as input;

the adaptive scheduling radius ACR is calculated as follows:

is a confidence interval size solving function, D _min Is the shortest distance to successfully locate the predicted coordinates;

wherein,,

D _min ＝max(dis(N[n ₁ ,n ₂ ,n ₃ ]))

where N [ ] is the set of scheduling nodes needed to successfully locate the predicted coordinates, and the function dis () is the distance between the calculated predicted coordinates and the node.

2. The adaptive physiological monitoring and intelligent scheduling positioning system according to claim 1, wherein the RM-MAC improvement protocol incorporates an indoor positioning rescue function into the physiological monitoring system by adding a rescue type data packet in a superframe structure of the MAC protocol;

the RS algorithm is a relay node selection algorithm designed for RM-MAC improvement protocol for adaptively selecting an appropriate data forwarding relay node according to the priority level of the data.

3. The adaptive physiological monitor and intelligent dispatch positioning system of claim 2, wherein the physiological data monitor module is comprised of WBANs, RM-MAC improvement protocols and RS algorithms comprised of a plurality of wearable wireless biosensors.

4. The adaptive physiological monitoring and intelligent scheduling positioning method based on the adaptive physiological monitoring and intelligent scheduling positioning system as claimed in claim 1, which is characterized by comprising the following steps: