Routing of Sensor networks

1. Routing characteristics of sensor networks

As a self-organizing dynamic network, the sensor network needs the network to heal automatically when the node fails and the new node joins, which causes the network topology to change dynamically. The key technology of the sensor network is the multi-hop self-Organizing Network routing protocol, in addition, the routing mechanism in the sensor network needs to be connected with the data fusion technology. According to the different sensor network application, there are four kinds of thought routing protocols. Energy-aware routing is a routing protocol that uses network energy efficiently, which focuses on the least energy consumption and the longest network lifetime, and so on, and the query-based routing is suitable for the application of environmental detection, in which the data collected by the sensor nodes are constantly queried. The traffic flow focuses on controlling commands and collecting information between the aggregation node and the sensor node. The data information collected by the sensor node usually needs data fusion, which can save energy by reducing the traffic flow, and the geo-routing protocol is suitable for applications such as target tracking, and it is necessary to wake the sensor node closest to the target when tracking the target, in order to obtain more precise location information about the target. The route selection is based on the location information of the node, and the reliable routing protocol is mainly to satisfy the reliable QoS of the routing technology.

2. Directional Diffusion Routing Technology

Directional diffusion Routing (Directed diffusion) DD is a routing technique based on the idea of query routing. It takes three stages to establish the route, namely the interest diffusion stage, the gradient establishment stage and the path strengthening stage. During the interest diffusion phase, the aggregation node broadcasts interest information to the neighbor nodes periodically. Interest information refers to the information that the aggregation node needs to query, which contains information such as task type, target area, data sending rate, etc. Each node that receives the interest information stores a list of interests locally, each of which records information such as who sent the interest and the sending rate timestamp, which is important and is the basis for establishing a gradient. Each table entry also has a field that represents a valid time value for this item, which will be deleted once the time has passed. If the node receives the same neighbor interest information as a table item in the local interest list, the node updates the valid time value in this table entry. If the interest information received is the same as the interest information that the node has just forwarded, in order to avoid entering the loop the node will discard this interest information. If the information received by the node does not match the local interest list successfully, a new table entry is added to the interest table.

Next is the gradient establishment phase, when the source node collects the task data specified in the interest information, it sends the data to the neighboring nodes on the gradient and sets the rate at which the data is collected based on the data send rate on the interest information. Because the source node may receive interest information from multiple neighbor nodes, the source node sends the data to each neighbor node in the list of interests. This allows the aggregation node to receive multiple different paths but with the same content of the data. For intermediate nodes, when it receives information, it first queries the table entries of the interest list and discards the data if there is no matching data. If there is a corresponding interest table entry, the data buffer pool corresponding to this interest is checked, and the data buffer pool is used to hold the most recently forwarded data. If there is a copy of the data in the buffer pool that matches the received data, that means that the data has been forwarded and discarded in order to avoid the transmission loop. Otherwise, if the sending rate of the neighbor node in the interest table is greater than the sent rate of the received information, the node forwards all the received data. If it is less than the sent rate of the received information, it will only be forwarded as part of the process, since the gradient setup phase starts with a selection of high-rate packets to discard packets with very low data rates, thus preparing for path enhancement. Forwarded packets are cached into the data buffer pool and the forwarding time is recorded.

In the interest diffusion stage, the data transmission path of the source node to the aggregation node is established, but at this time the rate of the source node is very low, and the rate of transmitting data is lower, and the gradient is the detection gradient. When the gradient-directed diffusion routing technique is established, the optimal path is established by the forward strengthening mechanism, and the gradient relation of data forwarding is modified dynamically according to the change of network topological structure. The optimal path is to extend from the aggregation node to the source node, and first the aggregation node chooses the neighbor node that first sends the new data as part of the optimal path, at which point the aggregation node sends a reinforcing message to the neighbor node. We know that interest information will contain a data rate option, and it will begin to work at this point. The data rate is included in the enhanced information, but it is a large data rate rather than a very low data rate. When the Neighbor node receives this information, it finds that it only increases the data rate, so it analyzes that this is a path-enhancing information and therefore updates the data rate in the interest table. The next neighbor node is similar to a convergence node repeating the above steps until the next hop is the source node, thus establishing an optimal transmission path from the source node to the aggregation node.

Of course, the criteria for path strengthening are not unique, but there are other options available in addition to the speed at which the data is sent. For example, you can choose a certain time to send the most data node or the most stable data transmission node, depending on the actual application needs to be determined, the most important thing is to establish the idea of routing. The rate of data transfer on this path will be greatly improved when the enhanced path is established, and if the rate of a node is suddenly found to be reduced then the node will re-select the reinforcing node. This path is not always used after this path is established, and the aggregation node will still periodically broadcast interest information to see if there is a better path. It can be seen that directional diffusion routing is a data-centric routing protocol, and the process of broadcasting a dynamic routing to a neighbor node after the aggregation node defines different interest messages according to the actual needs. The intermediate sensor node between the source node and the aggregation node through the interest table establishes multiple data transmission paths from the source node to the aggregation node. This routing mechanism also has drawbacks, because to flood interest packets, this will consume more time and energy.

Routing of Sensor networks