Application layer technology of sensor network

The basic purpose of sensor node function is sensing, detecting and sensing, the sensor node and the communication and networking technology in the sensor network together constitute a complete sensor network. But for the specific application, the basic technology of the application layer is needed to support the sensor network to complete the task, including time synchronization technology, location technology, data fusion technology and energy management technology.

1. Time Synchronization Technology

　　First we need to know why time synchronization technology is needed. For example, to test the speed of a car, you can set up 2 sensors on the road, according to the distance of the 2 sensors and the car through the time difference between the 2 sensors can calculate the speed of the car. The distance is fixed, the error is mainly reflected in the time difference, that is, when the car passes through the first node, the second node should be the same as the first node. In addition, there are many complex applications that require coordination and collaboration between multiple sensors. The time synchronization technology of the existing network often wants to reduce the time error to the extreme, but in the sensor network due to the power limit is difficult to complete the precise time synchronization technology, only in the error acceptable range on the line. At present, there is a kind of more mature sensor network time synchronization technology TPSN (Timing-sync Protocol for sensor Networks), it is still using layered thinking step by step to synchronize time to all nodes in the network. TPSN needs to have a root node, which can communicate with the outside world to get accurate time, it can assemble complex hardware such as GPS and other aspects of better configuration. The root node acts as the clock source in the sensor network, and we divide the root node into the No. 0 layer. The next part of the node is set to the 1th layer, the other part is set to the 2nd layer, and then can be divided into 3rd or 4th layer, specific details. This tpsn the entire network into a hierarchical structure. We call the hierarchical discovery phase, where the root node broadcasts the level Discovery packet package, the level discovery grouping contains the ID and level of the sending node, and the ID is an identity for each node. When a neighbor node of the root node receives a Discovery packet packet, it sets its own level to the level in the packet plus 1, which establishes level 1th. The 1th-level node continues to broadcast new discovery packet packets to its neighbor node, noting that this place ignores any other level of discovery grouping once the node determines the level. Next is the second stage, called the synchronization phase, where the root node now starts the broadcast time synchronization grouping. The 1th-level nodes each wait for a random time after they receive the synchronization packet, and this time node exchanges the synchronization message with the root node to achieve synchronization with the root node. After the second-level node hears the exchange message of the 1th level node, waits for a period of time after the 1th node completes the time synchronization, it starts and synchronizes the message exchange with the 1th level node. Finally, each node is synchronized with the closest previous node in the hierarchy to achieve synchronization of the entire network. Look at the text may be a bit dizzy, here is a time synchronization diagram I drew. Of course, this knowledge is the theoretical knowledge of textbooks, do not know what the actual development or application is like, and do not know what the actual communication problems.

To improve time accuracy, the TPSN protocol adds a time field to the synchronization message before the MAC layer message is sent out. The synchronization error of the TPSN protocol is proportional to the distance of the hop number, it is obvious that there is a time error between each level node and the next level node, and there is an error between the next node and the next level node. The TPSN protocol realizes the time synchronization of the nodes in the whole network, but this is a short time synchronization, because the sensor network is affected by the environmental factors will change dynamically, so if you need to synchronize long time, you can broadcast periodically. I can see that there are two common ideas in the sensor network, one is separation, all nodes are layered or divided, and one is cyclical.

2. Positioning technology

sensor node positioning is important, and when the sensor detects information, most of the information is where it is expected to occur. First of all, 2 concepts, sensor nodes can be divided into anchor points and unknown nodes, Anchor Point is sometimes called Beacon node, it occupies a small proportion of network nodes, you can configure GPS positioning to obtain their own precise position. The anchor point is the reference point for other unknown nodes to realize the positioning; The second concept is that we always say neighbor nodes, which refer to all other nodes within the radius of communication within the sensor node. In the Wireless LAN location is the reference point of the AP as a known coordinate, the mobile terminal receives the detection information from the AP and guesses the distance according to the intensity of the signal. But also to know that the sensor network is still very bitter, they are in the environment with a lot of uncertainty, but also a most headache is the problem is electrical energy, so about the sensor network related technologies require low power, fault tolerance can automatically correct errors, overcome external interference factors. On these basis for the positioning system need to consider 2 points, one point is the accuracy of positioning, a good refresh rate. Sensor network positioning technology is mainly divided into location-based positioning technology and no ranging location technology, the following are introduced.

The locating technique based on ranging is measured by the distance between nodes, and then the position of the node is obtained by mathematical calculation on the basis of multiple distances. Measurement distances can be obtained through RSSI or Toa/tdoa. Rssi refers to the receiving signal strength indication, which is measured by measuring the energy level of the receiving RF signal to determine the distance from the transmitting node, but the accuracy is not very high. Toa/tdoa refers to the arrival time/arrival timing technology, TOA and Tdoa This is 2 different technologies, but all through the transmission time to estimate the distance between two nodes, high precision, but because the wireless signal transmission speed as long as there is a little time error will cause great error. Toa is known for transmission speed and transmission time, which is based on these 2 conditions to get the distance. TDOA is the simultaneous transmission of two different speed signals, according to the two kinds of signal speed and the time of arrival to get distance. To get the distance after the multilateral positioning method and Min-max positioning method, multilateral positioning is to get multiple anchor points to a target node distance, after the distance can be anchored to the center of the circle, the intersection between these circles is the target node. The multilateral positioning solution equation will have floating point arithmetic, so the computational cost is high. Min-max positioning is based on the location of the anchor point to the target node to create a square instead of a circle, these squares can get a minimum range of rectangles, take the rectangle's centroid can be, it is relatively simple, but the positioning accuracy is not high-level positioning.

No ranging location technology is based on network connectivity to determine the number of hops between network nodes, by setting a approximate distance to the number of hops and then using the hop number multiplied by the hop distance. Obviously this method is not high precision, but its advantage is that the cost is small compared to energy saving. There are 2 kinds of algorithms, one is centroid algorithm, the anchor point will broadcast its own coordinate location information, so when the target node receives this information will be reduced to a shape to compare the rules of the polygon, and then use the anchor point in this polygon to calculate the centroid, that is, all the x-coordinate and then averaged, The same processing is done for the y-coordinate, and then the location information of the target node is so simple and rough. There is also a dv-hop algorithm, nor is it a very reliable algorithm. It calculates the approximate distance of one hop based on the distance between the anchor point and the anchor point and the hop count, and then uses a multi-position or Min-max to locate it.

3. Data fusion

Data fusion is very important in the sensor network, its essential function is 3 points: Save the energy of the network, improve the accuracy, improve the efficiency of data collection. Sensor network deployment is often a large number of delivery sensors to the target area, because the individual sensor monitoring capacity is limited so that the sensor to achieve a certain density to meet the demand, which will enhance the robustness of the entire network and accuracy. However, the drawbacks are obvious, and multiple sensor nodes may be more concentrated in a region, resulting in redundancy. This time data fusion can be in the middle node forwarding when the data deduplication, there have been experiments show that the sensor sends a bit of energy consumption is far greater than the energy consumed by executing an instruction, so that data fusion can reduce network consumption to a minimum, which is the role of energy saving; Due to the cost and volume constraints, the function of the node is often not very powerful, so a large number of nodes to detect, when the node is very large if all the information at the end of the aggregation node to summarize because there will be a lot of data is big error, resulting in the final result error. If the fusion in a small area, because the data between the nodes are similar, so the data with greater error can be deleted directly to improve accuracy; Internal data fusion can reduce data transmission, reduce network congestion, and reduce latency, so sensor nodes can spend more time collecting data to improve data collection efficiency.

Data fusion can be divided into loss-free fusion and loss fusion. Loss-free fusion refers to the fact that all the details are retained only to remove redundant portions of the information, it only reduces the length of the packet head and control overhead, the specific data will remain unchanged. For example, in real-time acquisition of a room temperature information, the report node received multiple data if the same data is only select the time the latest data to report, that is, only modify the head of the time information, and the data unchanged. Loss of fusion refers to discarding some detail information to reduce the amount of data that needs to be stored or transmitted, for example, to get the lowest temperature of a room, to take a minimum of the data received by a reporting node in a region, while other data is discarded. In addition there are many ways to categorize the way it is not introduced here.

4. Energy Management

The energy management of sensor network is mainly embodied in 2 aspects of power management and energy-saving communication protocol of sensor node. Sensor nodes typically have 4 parts: a processor unit, a wireless transmission unit, a sensor unit, and a power management unit. The energy consumption of the sensor unit is related to the complexity of the application, but its energy consumption is negligible compared to the wireless transmission unit. Processor loss and wireless transmission is a part of the need to consider power management, the processor part is mainly hardware improvements, I am concerned about the loss of the wireless transmission portion. The Wireless transmission section contains the entire transmission process, so you should reduce energy consumption as much as possible from the beginning of each layer protocol. The core part of the Sensor network protocol stack is the data link layer and the network layer. The data link layer controls the way that wireless channels are used between neighboring nodes, which determines the sending, receiving, listening, and sleeping state of the nodes. The use of listening/sleeping mechanism can be very good energy saving. The network layer is responsible for selecting the best route for data transmission, which consumes energy and increases the energy consumption as the communication distance grows. And when a node sends data to another node, the short-distance multi-hop approach consumes less energy than a single hop over long distances, so choosing the right route is very important. In the application layer, the above introduction is known to have data fusion for energy saving.

Application layer technology for sensor networks