CN109756276B

CN109756276B - High-reliability low-delay data transmission method in body area network

Info

Publication number: CN109756276B
Application number: CN201910080597.8A
Authority: CN
Inventors: 孙罡; 王凯; 孙健; 虞红芳
Original assignee: University of Electronic Science and Technology of China
Current assignee: University of Electronic Science and Technology of China
Priority date: 2019-01-28
Filing date: 2019-01-28
Publication date: 2020-03-10
Anticipated expiration: 2039-01-28
Also published as: CN109756276A

Abstract

The invention discloses a high-reliability low-delay data transmission method in a body area network, which comprises the following steps: s1, calculating the node sampling time of each node according to the number of the new and old data frames of the node and the channel model; s2, calculating the node data frame length of each node according to the currently allocated time slot segment and the channel model of the node; and S3, packaging the sampling data into a data frame according to the length of the node data frame at the node sampling moment, and sending the data frame to the local processing unit. The scheme for determining the length of the data frame consumes less time slots under the condition of transmitting the same amount of service data, so that the time slot utilization rate and the average delivery probability are higher, and the advantages are more obvious when the channel condition is poorer.

Description

High-reliability low-delay data transmission method in body area network

Technical Field

The invention relates to the technical field of body area networks, in particular to a high-reliability low-delay data transmission method in a body area network.

Background

The wireless body area network is a wireless sensor network which takes a human body as a center and consists of a local processing unit (intelligent terminals such as a mobile phone and a bracelet) and wireless sensors distributed on the surface of the human body or in the human body. Usually, the local processing unit and the common sensor node directly perform data transmission through single-hop communication to form a star network. The sensor nodes collect information of each part of the human body, and then transmit the information to the local processing unit for processing and displaying or further transmit the information to the remote processing center.

Due to the particularities of wireless body area networks themselves, the study of their communication protocols is facing several challenges. Firstly, the blocking and absorption of the electromagnetic wave signals by the human body can cause the deep fading of the channel of the wireless body area network, the deep fading time of the body surface channel is as long as 400ms, and the value is far higher than that of the traditional wireless sensor network. Secondly, the network environment and the service requirements of the wireless body area network are also in dynamic change due to the movement of the human body. For example, when a heart patient is moving violently, the data (such as an electrocardiogram) associated with the movement should have a lower transmission delay and a lower packet loss rate. In addition, since the sensor node is usually worn on the human body, and the size and battery capacity of the sensor node are very severely limited, the wireless body area network is an extremely energy-limited system. Finally, due to the weak processing power of the sensor nodes, generally speaking, complex computation tasks should be performed by the local processing unit, while the sensor nodes are only responsible for simple collection and transmission tasks.

The MAC protocols for wireless body area networks are generally divided into three categories: time division multiplexing based, contention based, and hybrid MAC protocols. The MAC protocol based on time division multiplexing divides time into time slots with the same size, and different nodes obtain different time slots. The node wakes up and transmits data when the time slot belonging to the node starts, and enters a sleep state in other time slots. This mechanism can avoid collisions and reduce energy consumption due to the introduction of the standby state, but puts high demands on the slot allocation strategy due to the dynamic variation of channel conditions and the diversity of traffic demands of different nodes. The MAC protocol based on competition does not distribute the resources uniformly, and the nodes adopt a certain mechanism to obtain the shared channel resources in a competition mode. This mechanism is very scalable, however, the simultaneous transmission of highly correlated data introduces a large number of collisions. Hybrid MAC protocols use both time division multiplexing and contention based mechanisms, but are often too complex.

At present, there have been some studies on MAC layer data transmission strategies. For example, the solution models the QoS constraints and energy constraints of the sensor nodes, gives a definition of the cost, and proposes a slot allocation and transmission power and transmission rate adjustment solution to minimize the total cost of all nodes. But the protocol interaction process has disadvantages, which results in longer waiting time delay of the node and low time delay performance of the system.

Researchers model the priority and utility functions of the sensor nodes, and allocate transmission time slot segments with better channel conditions and more time slot numbers to nodes with higher priority so as to maximize the total utility of all the nodes. Although the method can dynamically adjust the time slot allocation according to the change of the channel state to obtain a larger utility value, the low-priority node usually needs to consume a plurality of time slots to successfully transmit the data frame due to the poor channel condition, and the total delivery probability and the time slot utilization rate are not high.

Disclosure of Invention

Aiming at the defects in the prior art, the high-reliability low-delay data transmission method in the body area network provided by the invention solves the problem of higher total cost of all nodes.

In order to achieve the purpose of the invention, the invention adopts the technical scheme that: a high-reliability low-delay data transmission method in a body area network comprises the following steps:

s1, calculating the node sampling time of each node according to the number of the new and old data frames of the node and the channel model;

s2, calculating the node data frame length of each node according to the currently allocated time slot segment and the channel model of the node;

and S3, packaging the sampling data into a data frame according to the length of the node data frame at the node sampling moment, and sending the data frame to the local processing unit.

Further: the calculation method of the node sampling time in the step S1 is as follows:

s11, calculating the average delivery probability of the new data frames and the average delivery probability of the old data frames according to the number of the new and old data frames of the node and the channel model;

s12, when the node sampling rate is not less than the data frame sending rate, the step S13 is executed, otherwise, the step S14 is executed;

s13, calculating the node sampling time according to the average delivery probability of the old data frame, and entering the step S2;

s14, the node sampling time is calculated from the average delivery probability of the new data frame, and the process proceeds to step S2.

Further: the calculation formula of the node sampling time in step S13 is as follows:

in the above formula, t_start,sWhen sampling for a nodeMoment t_startStarting time of time slot segment allocated for node, N_oldFor the number of old data frames in the buffer, R_frameFor the rate at which the data frames are sent,

is the average delivery probability of the old data frame.

Further: the calculation formula of the node sampling time in step S14 is as follows:

in the above formula, t_start,sFor the sampling time of the node, t_startStarting time of time slot segment allocated for node, N_slotsTotal number of time slots, T, allocated to node i_slotIs the length of the time slot, T_sIn order to be the time of sampling,

is the average delivery probability of a new data frame.

Further: the method for calculating the length of the node data frame in step S2 includes:

s21, calculating the current average delivery probability of the node according to the currently allocated time slot segment and the channel model of the node;

s22, calculating the optimal data frame segmentation segment number through the average delivery probability;

and S23, calculating the length of the node data frame according to the optimal data frame segmentation number.

Further: the calculation formula of the optimal data frame segmentation number in step S22 is as follows:

in the above formula, N_partsSegmenting the number of segments, P, for an optimal data frame_del,frame,kIs the average delivery probability.

Further: the calculation formula of the node data frame length in step S23 is as follows:

in the above formula, L_frame,shortFor node data frame length, L_frame,longFor the node original data frame length, L_ohFor node overhead bit length, N_partsThe number of segments is sliced for the optimal data frame.

The invention has the beneficial effects that:

(1) the time delay characteristic is improved. The invention provides that the node wakes up at a time point near the start of the distributed transmission time to sample, and gives a very accurate calculation formula of the sampling start time by comprehensively considering the requirements of throughput and the like, so that the data generated by sampling can directly package data frames and enter a buffer queue, the waiting time delay is greatly reduced, and the time delay performance is further improved.

(2) The average delivery probability is high. The scheme for determining the length of the data frame consumes less time slots under the condition of transmitting the same amount of service data, so that the time slot utilization rate and the average delivery probability are higher, and the advantages are more obvious when the channel condition is poorer.

(3) The energy efficiency is higher. The determination scheme of the data frame length provided by the invention consumes energy under the condition of transmitting the same amount of service data, so the overall energy efficiency is higher, and the advantage is more obvious when the channel condition is poorer and the energy is insufficient.

Drawings

FIG. 1 is a general flow chart of the present invention;

FIG. 2 is a flowchart of step S1 according to the present invention;

FIG. 3 is a graph of the delivery probability p (k) versus time slot k for a good initial channel state in accordance with the present invention;

FIG. 4 is a graph of the delivery probability p (k) versus time slot k for bad initial channel state in the present invention;

FIG. 5 is a graph showing the ratio of the number of consumed time slots of two data frame lengths under different average delivery probabilities according to the variation of the number of data frame segmentation segments;

FIG. 6 is a graph showing the gap ratio between two data frames consumed at an average delivery probability of 0.35 according to the number of data frame segmentation segments;

FIG. 7 is a graph of the number of optimal data frame slices as a function of average rendering probability in accordance with the present invention;

FIG. 8 is a graph of the optimal consumed timeslot number ratio as a function of average delivery probability in accordance with the present invention;

fig. 9 is a graph of the ratio of the total number of bits transmitted versus the average delivery probability in the present invention.

Detailed Description

The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.

According to the experiment and demonstration of the IEEE802.15.6 working group, the Markov model can well depict the channel characteristics. The invention adopts the two-state Markov process proposed in the existing research to model the channel between the sensor node and the local processing unit. The channel is divided into two states of good and bad, if the data frame can be successfully transmitted to the local processing unit under a certain channel state, the channel state is good, otherwise, the channel state is bad. The probability that a state becomes a good state after k slots (i.e., the delivery probability of a data frame in that state) is:

wherein, P_BGIndicating that the channel status has changed from bad to good, i.e., the probability that a data frame failed to be transmitted in the previous slot but was successfully transmitted in the current slot. P_ch＝P_BG+P_GB. From the expression, when p (0) ═ 0, that is, the initial state is bad, the delivery probability p (k) is with respect to k sheetsThe adjustment is increased, when p (0) is 1, the delivery probability p (k) is monotonically decreased with respect to k, and the delivery probability finally converges to a constant value

The change rule of p (k) with respect to k is shown in fig. 3 and 4.

It should be noted that the same node has different transition probability matrixes (i.e., P) under different human motion states or different channel conditions_BGAnd P_GB) However, in the same motion, the transition probability matrices are the same. Therefore, in the using process, for each specific activity, the local processing unit only needs to use the measured transition probability parameter for channel estimation.

The related behaviors of node sampling and data transmission in the existing research protocol interaction process are divided into the following steps:

(1) the node receives the beacon frame sent by the local processing unit at the beginning of each superframe and synchronizes with the local processing unit

(2) Sampling according to the sampling rate distributed by the local processing unit, storing the sampled data in a buffer queue, and then making the node sleep

(3) The node wakes up at the beginning of the transmission slot segment scheduled for it and sends the data in the buffer on a first come first served basis.

According to the protocol interaction process, the waiting time of the data frame is delayed by D_waitThe method is divided into two parts:

D_wait＝D_sleep+D_que

wherein D is_sleepRepresenting the sleep delay of a data frame from entering a buffer queue to waiting for the start of transmission of the corresponding node, D_queIndicating the queuing delay of the data frame waiting for the completion of the transmission of the data frame ahead of the data frame in the buffer queue. The data frame in the buffer queue is sent by adopting a first-come first-serve principle, so the queuing delay D_queCan be expressed as:

wherein N is_inBufIndicating the number of data frames in the buffer queue before the data frame at the current moment,

representing the average delivery probability during the transmission of these data frames, which can be determined by a channel model, R_frameIndicates the data transmission rate (unit of data frame, assuming that the length of the data frame is a fixed value). Since the channels transmit data under different channel environments with different packet loss rates, the average delivery probability corresponding to the time slots is used to more accurately describe the total time of the transmission.

Then for sleep delay D_sleepAnd (6) performing calculation. Assuming that the transmission start time divided by the node i is the jth time slot of the super frame, the sleep delay of all data frames of the node i can be expressed as:

D_sleep＝j·T_slot

it can be seen that, since the sensor node starts sampling after synchronization and then goes to sleep until the transmission time starts, the data frame generated by sampling is stored in the buffer until the allocated transmission time starts, and the sleep time delay D is_sleepAnd will also increase significantly. Therefore, the invention finely adjusts the interaction process between the sensor and the local processing unit specified by the protocol to reduce the sleep delay and improve the delay performance. Specifically, the node is regulated to wake up at a time point near the beginning of the distributed transmission time to sample, so that data generated by sampling can be directly packaged into a data frame and enter a buffer queue, sleep time delay is greatly reduced, and time delay performance is improved.

Although the sampling start time has been determined to be a time point near the transmission start time, the specific sampling time is also limited by many factors, and the present invention analyzes the constraints in the above process to determine the specific time at which sampling starts. On the one hand, if the sampling rate f is greater than the transmission rate R_frameThen there may be a sufficient number of data frames per time instantSo that transmission is obtained, i.e. throughput can be guaranteed; however, if the sampling time is too early, the average queuing delay of the data frames of the node will be increased, and considering that there is an old data frame which fails to be transmitted in the previous superframe in the buffer, the buffer may overflow, so that it is necessary to select a later sampling time, but at least it is necessary to ensure that the sampling is completed within the time slot segment allocated by the node, so as to avoid affecting the normal data transmission of the node. On the other hand, if the sampling rate f is less than the transmission rate R_frameThe sampling needs to be advanced to meet the throughput condition, so the sampling time needs to be advanced, and similarly, the buffer overflow is prevented.

As shown in fig. 1, a high-reliability low-latency data transmission method in a body area network includes the following steps:

as shown in fig. 2, the method for calculating the node sampling time includes:

the calculation formula of the node sampling time is as follows:

in the above formula, t_start,sFor the sampling time of the node, t_startStarting time of time slot segment allocated for node, N_oldFor the number of old data frames in the buffer, R_frameFor the rate at which the data frames are sent,

for old data framesAverage delivery probability. As can be seen from the above formula, in order to ensure throughput, it is assumed that sampling of a new data frame is started after an old data frame in a buffer is sent, so that a lower queuing delay can be ensured to the maximum extent; and f is satisfied between the sampling rate and the sending rate>R_frameTherefore, the throughput in the transmission process can be ensured; in addition, in order to ensure that the sampling process of the node can be completely completed in the allocated time slot section, the time slot section divided by the node is assumed to be (t)_start,N_slots) The node sampling time is T_s。

I.e. the sum of the total time taken to transmit an old data frame and the sampling time is smaller than the total slot length divided by the data frame.

Generally, under the premise of considering the packet loss rate, the traffic requirements of all nodes (including the old data frames which are not transmitted and the new data frames generated by sampling) should be satisfied except for a few nodes with lower priorities.

Wherein N is_slotsRepresenting the number of all time slots (including the new and old data frames) divided by the node, the two equations are left-hand differenced:

since f is not less than R_frameIt is apparent that Δ ≧ 0, i.e.:

If f < R_frameNeed to be ensuredAnd ensuring that the queuing delay is small under the conditions of no buffer overflow and high throughput, namely setting an early sampling start moment. If sampling is started after the transmission of the old data frame is completed, since f < R_frameThen, the sampling of data cannot keep up with the sending rate, resulting in a decrease in channel utilization, a decrease in throughput, and even a failure to complete the transmission of data frames in the divided specified time slot segment. It follows, therefore, that the sampling start time must be before the transmission of the old data frame is completed. Here, an expression of sampling start time is given, and a calculation formula of node sampling time is as follows:

is the average delivery probability of a new data frame. Assuming that the new data frame generated by sampling is just completely transmitted successfully when all the time slot segments allocated by the node are ended, the sampling start time is set as the difference value between the right end point of the divided time slot segments and the sampling time. Since a certain packet loss rate exists in the new data frame generated in the period of time during the transmission process, the quotient of the sampling time and the average delivery probability of the corresponding period of time is used to more accurately represent the sampling time under the condition of considering the packet loss. The advantages of using the above equation to determine the sampling start time are the following: on one hand, the sampling starting time of the node is set to a later time so as to reduce the queuing time delay of the data frame as much as possible; on the other hand, since the right boundary of the divided slot segment is considered, the data frame can be transmitted in the predetermined slot segment. In addition, in order to satisfy the throughput condition, the sampling start timing needs to satisfy the following condition:

the sampling time is differed from the time when the transmission of the old data frame is completed by:

wherein, since only one data frame is supposed to be transmitted in one time slot, the transmission rate R and the time slot length T_slotSatisfy R_frame·T _slot1. From f < R_frameIt is found that Δ is < 0.

S2, calculating the node data frame length of each node according to the currently allocated time slot segment and the channel model of the node; the method for calculating the length of the node data frame comprises the following steps:

the time delay performance of the system is improved to a certain extent by the design of the sampling time of the nodes, and the indexes such as delivery probability, energy efficiency and the like are optimized by adjusting the length of a data frame in the data transmission process, so that the overall performance of the system is improved. Firstly, the influence of the length of the data frame on various indexes in the transmission process is analyzed. Due to the introduction of the retransmission mechanism, if a data frame fails to be transmitted in the current time slot, the data frame continues to be transmitted in the next time slot. More importantly, even if only a small portion of the data in the data frame fails to be transmitted, all of the data in the data frame is retransmitted. Intuitively, under the condition that other conditions are not changed, the larger the length of the data frame is, the higher the probability of transmission failure of the data frame is, and the higher the retransmission cost is. Assuming that the bit errors of each bit are independent during the transmission of a data frame, the delivery probability of the data frame can be expressed as:

P_del,frame(γ,L)＝(1-P_b,B(γ))^L

wherein L represents the bit length of the data frame, P_b,BWhich shows the equivalent bit error rate when a Differential Phase Shift Keying (DPSK) modulation scheme and a BCH coding scheme are used. The body area network official standard ieee802.15.6 specifies that DPSK differential phase shift keying is used for modulation in the ISM band and BCH is used for channel coding. γ represents the bit signal-to-noise ratio, which can be expressed as:

wherein, P_tx,dBRepresenting the transmission power (in dB), P_NIs the noise power, B is the system bandwidth, and R is the transmission rate. Pl (d) represents the path loss, which can be expressed as:

where d denotes the distance between the sensor and the local processing unit,

indicates the reference distance d₀Lower path loss; n represents a path loss exponent having different values for line-of-sight and non-line-of-sight channels; x_σRepresenting shadow effect, following a normal distribution, as a function of body movement, X_σThe mean and standard deviation of (a) will also vary.

The delivery probability of a single data frame in a specific channel environment, considering that the channel is dynamically changed with time, gives an expression of the average delivery probability of the data frame:

wherein, P (gamma | mu)_γdB，σ_γdB) Probability density function representing bit signal-to-noise ratio, subject to normal logarithmic distribution, X_σ·μ_γdBIs the mean value of gamma, sigma_γdBIs its standard deviation. In addition, as can be seen from the results of the prior studies, the average delivery probability of the data frame is bitMean μ of the signal-to-noise ratio_γdBIs a monotonically increasing function of.

In the case where the bit signal-to-noise ratio γ is constant, the delivery probability of a data frame increases as its length L increases, and therefore, the delivery probability of each data frame can be increased by decreasing the length of the data frame. One possible solution is to change the length of the variable field in each data frame, i.e. to divide the data originally transmitted by one data frame into several data frames for transmission, so that the actual number of time slots required for successful transmission of each data frame is reduced, and the total time slots required for transmitting all data is also reduced. However, since each data frame includes fixed overhead bits such as MAC frame header information, frame check sequence, etc., reducing the length of the data frame means increasing the total overhead bits, and without considering the delivery probability, the total traffic demand of the node may increase, and more time slots may be needed to transmit a fixed amount of traffic data, while consuming more energy.

As can be seen from the above analysis, reducing the length of the data frame results in two different trends of increasing and decreasing the number of the required total time slots, so that a quantitative evaluation needs to be performed on the total data amount sent during the process of sending the data frame by each data frame to determine the feasibility of the scheme. The subscripts long and short are used herein to distinguish between long and short data frame related parameters, respectively. Assume the original data frame length L_frame,longThe total service requirement of the lower node is Req_frame,long(data frame unit, including overhead bits), the average delivery probability of a node transmitting data in a divided time slot is

The total data bit number theoretically required to be transmitted in the process of the node completing all service requirements is as follows:

the total data bit number transmitted by the node is more accurately described by taking the packet loss rate in the transmission process into consideration. In addition, it should be noted that the total number of bits transmitted can be used to describe the energy consumption of the node. Assuming that the transmission rates used for the acknowledgment frame and the data frame are the same, the total number of slots consumed in this process is:

then, the correlation index after the length of the data frame is reduced is calculated. It is assumed here that the node will have originally demanded traffic in one data frame (excluding the overhead bit L)_oh) Average cut to N_partsEach data frame is transmitted, and since each data frame has fixed-size overhead bits, the length L of the new data frame_frame,shortAnd new business requirements Req_frame,short(regardless of delivery probability) are:

Req_frame,short＝N_parts·Req_frame,long

a shorter data frame length L can be derived_frame,shortThe following relevant indicators. Total data bit number N theoretically required to be transmitted when node completes all service demands_{tran,bit,short}Comprises the following steps:

the total number of time slots N consumed in the process_{slots,tran,short}Comprises the following steps:

firstly, analyzing the total time slot number theoretically needed by the node to complete all service requirements:

wherein

The ratio of the average delivery probability of the two is expressed, and the increase of the average delivery probability caused by reducing the length of the data frame is reflected;

slicing coefficient N representing a data frame_parts. When in use

I.e. the average delivery probability due to the reduced data frame length increases by more than N_partsWhen the time is doubled, the number of the screws is doubled,

namely N_{slots,tran,short}<N_{slots,tran,long}At this time, the time slots consumed when the short data frames are used to complete the same amount of service demands are less, the short data frames should be used for data transmission, otherwise, the length of the data frames should be kept unchanged. However, cannot be directly derived

And 1, and therefore further analysis is required. Due to the fact that

The delivery probability in (1) represents the average delivery probability of all data frames transmitted by the node under different lengths of the data frames, and the calculation complexity is high, so that the average delivery probability of all data frames is approximated to the delivery probability P of one data frame (assumed to be the kth data frame in the service requirement) in the transmission process_del,frame,kThe above formula is simplified as follows:

L_ohand L_frame,longIs constant, so ratio_slots,tranIs P_del,frame,kAnd N_partsAs a function of (c). The invention adopts an experimental mode to carry out the delivery probability P_del,frame,kThe influence of the number of segments of the next different data frame on the number of consumed time slots is simulated, as shown in fig. 5. In the figure, the overhead bit size is 9 bits according to the specification of IEEE802.15.6, and the length L of the original data frame_frame,longContains a variable field, here taken on average, of about 360 bits. The average delivery probability is taken from the set {0.025,0.05,0.075,0.1,0.15, 0.2, 0.25, 0.3}, and is represented by fig. 5(a) and 5(b), respectively, and is selected by taking the following factors into consideration:

since when P is present_{del，frame,k}>Ratio of ratio at 0.3_slots,tranAll values of (a) are greater than 1, that is, under the condition of considering delivery probability, the same amount of service data is transmitted, and no matter how small the original data frame is segmented, more time slots are consumed, for example, when P is used_{del，frame，k}The relationship between the ratio of the number of consumed slots and the number of segments of the data frame is shown in fig. 6 when it is 0.35, and therefore the delivery probability value in this range is not considered here.

Due to P_del,frame,kThe advantage of using short data frames in terms of the number of consumed time slots is more evident when smaller, so here we bound 0.1, the parts larger than 0.1 take 4 values at intervals of 0.05, and the parts smaller than or equal to 0.1 take 4 values at intervals of 0.005.

As shown in fig. 5, under each delivery probability within the range, the ratio of the number of consumed time slots of the two data frame lengths is increased after being decreased. This is because, with the increase of the number of segmentation segments, the delivery probability of a single data frame under the short data frame length is higher and higher, and the increase thereof is far larger than the increase of the total overhead bits caused by the decrease of the data frame length, at this time, the total number of time slots consumed by using the short data frame is smaller, which means that the time slot utilization rate and the total effective average delivery probability are higher on the premise of no change of the service requirement. When the number of segmentation segments is increased to a certain degree, the increase of the number of consumed time slots caused by the increase of the number of total overhead bits gradually offsets the increase caused by the increase of the delivery probability, and at the moment, the ratio shows an ascending trend until the ratio is more than 1.

With the increase of the delivery probability, the optimal number of segmentation segments of the data frame (i.e., the number of segmentation segments corresponding to the time slot ratio in each sub-graph when the time slot ratio is minimum) shows a downward trend, as shown in fig. 7. This is because when the delivery probability increases, the increase in delivery probability due to the reduction in the length of the data frame is relatively small, and if the number of segmentation segments increases, more overhead is introduced, further reducing the advantages brought by the short data frame.

As the delivery probability increases, the optimal slot number ratio of the data frame (i.e. the minimum value obtained by the slot number ratio in each sub-graph) shows an upward trend, which means that the gain brought by the short data frame is decreased, as shown in fig. 8. Similarly, this is because as the delivery probability increases, the increase in delivery probability by decreasing the length of the data frame becomes relatively smaller and smaller, and thus the total number of consumed time slots also increases. When the delivery probability is small, the ratio of the consumed time slot number is far less than 1, and the time slot consumption can be reduced by more than 70% by reducing the length of the data frame.

Fitting the curve in fig. 7 can obtain the relationship between the optimal time slot segmentation number and the average delivery probability, that is, the calculation formula of the optimal data frame segmentation number is as follows:

From the above analysis, the average delivery probability P_{del，frame，k}When the number of the data frames is less than or equal to 0.3, the time slot consumption can be greatly reduced under the condition of finishing the equal service requirement by reducing the length of the data frames according to the formula, and the time slot utilization rate and the average delivery probability are improved. However, since the derivation process does not consider the magnitude relationship of the total number of bits transmitted (i.e. energy efficiency) at different data frame lengths, the influence of the given relationship of the formula on energy efficiency needs to be further processedAnd (4) analyzing in one step.

The calculation formula of the node data frame length is as follows:

As shown in FIG. 9, the delivery probability is within the range of values, P_{del，frame，k}When the ratio is less than or equal to 0.3, ratio_tran,bitThe values of (a) are all smaller than 1, that is, under the condition of considering the delivery probability, the total number of bits actually transmitted when the same amount of service needs are completed by using the short data frame is small, and obviously, the total energy consumed at this time is also small. In addition, when the delivery probability is large, relatively much energy is consumed, because the improvement of the delivery probability due to the reduction of the data frame length is relatively small, and in addition, the overhead bits increased compared with the original data frame length are added, so that the total number of bits transmitted is increased, and the energy consumption is increased correspondingly.

From a combination of FIGS. 8 and 9 and corresponding analysis, it can be seen that at P_{del，frame，k}When the length of the node data frame is less than or equal to 0.3, the optimal average delivery probability, the time slot utilization rate and the energy efficiency can be obtained only by adjusting the length of the node data frame, the advantage is particularly obvious when the average delivery probability of the node under the length of the original data frame is small, for example, when the delivery probability is 0.05, the optimal segmentation section number is adopted for segmenting the data frame, and the equal amount of service data after transmission only needs about 40% of time slot number and about 15% of energy.

On one hand, when the channel condition is bad, such as the noise is too large, the bit signal to noise ratio is reduced, and the average delivery probability is reduced; on the other hand, if the node is not sufficiently powered and the transmission power has to be reduced, the average delivery probability is also reduced. Accordingly, the slot utilization and energy efficiency during transmission may be reduced.

No matter the channel condition is bad or the transmission power is insufficient due to the excessive noise, the bit signal-to-noise ratio in the transmission process is directly influenced, so that the average delivery probability is reduced, and the normal communication between the sensor node and the local processing unit is seriously influenced. The design of the data frame length is very suitable for the situation, and for the nodes with better channel conditions, the existing research scheme is adopted for data transmission, so that higher average delivery probability and energy efficiency can be obtained; for partial nodes with poor channel conditions and insufficient energy, the invention provides a data frame length adjustment strategy, and different data frame lengths are assigned to the nodes according to the average delivery probability of the nodes under the current channel conditions, so that the average delivery probability, the time slot utilization rate and the energy efficiency of the nodes are improved.

The data transmission strategy provided by the invention can be deployed in a wireless body area network MAC layer adopting single-hop communication in the scenes of military affairs, medical treatment, sports, entertainment and the like. The local body area network environment is composed of sensor equipment worn on the body surface or implanted in the body and a local processing unit (intelligent equipment such as a mobile phone and a tablet). The sensor device generally comprises three structural units, namely a radio frequency module, a sensor module and a storage module. The sensor module converts the energy form into an analog electric signal, and the analog electric signal is filtered by a filter and digitized by a data converter. The storage module is used for storing messages, and the radio frequency module is mainly used for receiving and transmitting signals. In addition, the sensor device and the local processing unit have corresponding software platforms for function control and information display.

The data transmission strategy provided by the invention is written into a corresponding platform in a software form, and then behaviors of collecting, transmitting data, receiving a confirmation frame and the like of the sensor are controlled.

The sensor equipment collects human body related data and maintains a local message cache queue, then the sampling time of the node is determined according to the design of the sampling time provided by the invention, the sampled data is packaged into a data frame according to the determination scheme of the length of the data frame provided by the invention, and the data frame is sent to the local processing unit and receives a confirmation frame message. After receiving the information of the sensor equipment, the local processing unit performs unified scheduling allocation on the time slots divided by the next superframe of all the sensor nodes, and then encapsulates the transmission plan and the sampling rate of the next superframe into a beacon frame and broadcasts the beacon frame to the sensor equipment. In the process, the data frame, the beacon frame and the confirmation frame all adopt the format specified by the IEEE802.15.6 official protocol.

After the local processing unit collects the data of the sensor device, it will perform a preliminary process, for example, sense the current channel state change according to the environment cognitive algorithm, determine the channel model parameters, and merge the data collected by different sensors according to their correlations. In addition, for more important data or abnormal data, the local processing unit uploads the data to the remote server for further processing, and the process can use various network communication modes such as WSN, WPAN, WLAN, internet and cellular network.

Claims

1. A high-reliability low-delay data transmission method in a body area network is characterized by comprising the following steps:

s1, calculating the node sampling time of each node according to the number of the new and old data frames of the node and the channel model, specifically:

s13, calculating the node sampling time according to the average delivery probability of the old data frame, and entering the step S2; the calculation formula of the node sampling time is as follows:

average delivery probability for old data frames;

s14, calculating the node sampling time according to the average delivery probability of the new data frame, and entering the step S2; the calculation formula of the node sampling time is as follows:

the average delivery probability of the new data frame;

s2, calculating the node data frame length of each node according to the currently allocated time slot segment and the channel model, specifically:

s23, calculating the length of the node data frame according to the optimal data frame segmentation segment number;

2. The method for transmitting high-reliability low-latency data in a body area network according to claim 1, wherein the calculation formula of the optimal number of segments of the data frame in step S22 is as follows:

3. The method for high-reliability low-latency data transmission in a body area network according to claim 1, wherein the calculation formula of the node data frame length in step S23 is as follows: