1 Introduction
-
Different data arrival patterns under various scenarios: Data in routine health monitoring are generated periodically where the data arrival rate can be determined by the sampling rate uniquely. But when the collected data exceed the safe range, sensor nodes generate burst data or emergency data that should be processed timely. What’s more, additional data are collected to improve the accuracy of burst traffic. The data arrival pattern is hard to predict.
-
Heterogeneous data arrival rates: Disparate sensor nodes are different in traffic varieties. Sampling text information, images, and video information are in different arrival rates for sensors. For example, electrocardiogram data rate is multiple times higher in comparison with body temperature.
-
Correlation between sensor nodes: Sensors in health monitoring are correlated with each other. A patient suffering from a fever will trigger the changes in temperature, blood pressure, and respiration sensors at the same time [13]. If all the related data are packed or clumped together to be transmitted in disorder, the strongest correlation data of vital signs to the illness may endure a delay and make no sense in the worst-case scenario.
-
TTR MAC discriminates the monitoring data to periodic data and burst data. It conducts two-round reservations to guarantee the data transmissions for the two data types to reduce energy consumption and overhead induced by control frames.
-
We build a relationship between the appropriate number of allocation slots and heterogeneous data arrival rates, which helps to utilize the time resource more efficiently and save more energy.
-
In TTR MAC protocol, the hub uses a token to assign additional allocation intervals for burst data according to the user priority of nodes and the criticality of burst data. Because the correlation between sensor nodes may generate several burst data streams at the same time. In this way, TTR MAC protocol can guarantee the most emergency data to be transmitted effectively and timely.
-
A bit sequence-scheduled mechanism is designed for m-periodic (m>1) data, which expands the network capacity and better satisfies the QoS of nodes.
2 Related work
2.1 Motivation
-
How to transmit the periodic and burst data generated in health monitoring effectively?
-
How to assign appropriate allocation intervals for the sensor nodes with diverging data arrival rates?
-
How to guarantee lower delay for nodes with high user priority and critical conditions when several burst data streams are generated due to the correlation between sensor nodes?
2.2 IEEE 802.15.6 overview
3 Proposed scheme
3.1 Data classification
3.2 Two-round reservation
3.2.1 Superframe structure
3.2.2 First-round reservation
3.2.3 Second-round reservation
-
User priority (UP)×IEEE 802.15.6 assigns different UPs for different frame types, which vary from UP0 to UP7. The UP is determined based on the data type generated by nodes, such as background data, video data, and emergency reports. The higher UP of a node is, the greater privilege to access the medium it has. UP7 is the highest while UP0 is the lowest. The predefined relationship of prioritized access can be attained in Table 1.Table 1User priorities representationUser prioritiesTraffic designationFrame type0BackgroundData1Best effortData2Excellent effortData3VideoData4VoiceData or management5Medical data or network controlData or management6High priority medical dataData or managementor network control7Emergency or medical implantedDataevent report
-
Health severity index (H)Health severity index denotes the degree of data deviating from the safety range, which is calculated by Eq. (1).$$ H = \left\{ \begin{array}{l} \min \left(\frac{{{\theta_{l}} - \theta }}{{{\theta_{u}} - {\theta_{l}}}},1\right) \;\;\;\;\theta < {\theta_{l}}\\ 0\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;{\theta_{l}} \le \theta \le {\theta_{u}}\\ \min \left(\frac{{\theta - {\theta_{u}}}}{{{\theta_{u}} - {\theta_{l}}}},1\right) \;\;\;\;\theta > {\theta_{u}} \end{array} \right., $$(1)where θ denotes the value of collected data, θ l denotes the lower bound, and θ u is the upper bound of the safe range. It can be seen that 0≤H≤1. The higher H is, the severer condition node is in.
-
Volume of burst data N BurstDataNodes estimate the volume of burst data in buffer in order to provide a reference for the hub to assign additional allocation slots for them. When the hub receives the second-round allocation request frame, the hub replies ACK with the added time field to inform the nodes to receive the second-round allocation assignment frame. Then, the hub will introduce the token to determine the transmission order of nodes with burst data. The token is calculated by considering both the node’s user priority and the critical condition.$$ \begin{array}{l} \;\;\;\;\;\;\;\text{token} = {\rho_{1}} \times \frac{{\text{UP}}}{8} + {\rho_{2}} \times H \\ \mathrm{s.t.}\;\;0 \le {\rho_{1}},{\rho_{2}} \le 1,{\rho_{1}} + {\rho_{2}} = 1 \end{array} $$(2)where ρ 1 is the weight factor of the user priority while ρ 2 is the weight factor of the critical condition to attain the token value.Nodes with the higher token are awarded with the higher privilege to transmit, which guarantees a better quality of service for the urgent data. Then, the hub assigns additional allocation slots for the burst data and writes the allocation interval in the second-round allocation assignment frame. It should also be emphasized that if the remaining allocation slots are insufficient to transmit all the burst data, the hub will deny to assign allocation slots for nodes with lower token.If the sensor nodes in health monitoring are sensitive to security, the hub will transmit the second-round allocation assignment frame in unicast manner, if not, in broadcast manner. The format of the second-round allocation assignment frame is the same as the first-round allocation assignment frame, as shown in Fig. 6 b. The difference is that there may be several allocation assignment frames following the length field in broadcast manner, each of which dedicates the start and the end of allocation intervals for the specific node.Upon receiving the second-round allocation assignment frame, nodes check whether the hub assigns additional slots for it or not. If it has assigned, the nodes will transmit the burst data when the specific allocation interval commences. Otherwise, it will wait for the next superframe to retry the second-round reservation.It should be noticed that the second-round reservation only takes place when node has burst data. If there is no burst data, all nodes and hub enter the sleep period to save energy.
3.2.4 An example of TTR MAC
3.3 Appropriate number of allocation slots
3.3.1 Allocation interval for nodes with specific data arrival rate
3.3.2 Number of allocation slots
4 Bit sequence mechanism for m-periodic data
4.1 Sequence function
4.2 Bit sequence-scheduled algorithm
5 Performance evaluation
5.1 Simulation parameters
Parameters | Value |
---|---|
pAllocationSlotMin | 500 μs |
pAllocationResolution | 500 μs |
pSIFS | 75 μs |
pExtraIFS | 10 μs |
mClockResolution | 4 μs |
Code rate (k
n) | 51/63 |
PSDU spreading factor | 1 |
Symbol rate (ksps) | 600 |
PLCP header spreading factor | 4 |
Voltage supply | 3 V |
Transmitted current (I
tx) | 7.5 mA |
Received current (I
rx) | 13.1 mA |
Standby current (I
idle) | 26 μs |
Sleep current (I
sleep) | 900 nA |