A hierarchical data acquisition method applied to a marine information network, comprising the following steps:
- step S1:
- arranging and anchoring a plurality of sensor nodes in clusters on a seafloor, wherein each of the clusters comprises a cluster head node and a plurality of ordinary nodes, the plurality of ordinary nodes acquire data information of the seafloor and transmit the data information to the cluster head node to aggregate the data information;
- step S2:
- releasing an autonomous underwater vehicle from a position of a water surface base station, and controlling the autonomous underwater vehicle to traverse the cluster head node of each of the clusters to acquire the aggregated data information and then to drive back to the water surface base station;
- step S3:
- controlling the autonomous underwater vehicle to transmit the acquired data information of the seafloor to the water surface base station;
- and
- step S4:
- after each of data acquisition periods, determining a sensor node in each of the clusters as a cluster head node in a next data acquisition period;
- wherein the arranging a plurality of sensor nodes in clusters in step S1 comprises the following steps:
- step S11:
- in cases that the plurality of sensor nodes are divided into 1 to n clusters, calculating, for each of the 1 to n clusters, a sum of squared errors (SSE) of sensor nodes to a cluster center, wherein the number of sensor nodes comprised in each of the clusters is random, and the sum of squared errors (SSE) is calculated by using the following equation:
- SSE
- =
- ∑
- i
- =
- 1
- k
-
- ∑
- x
- ∈
- C
- i
-
-
- x
- -
- x
- i
- _
-
- 2
- 2
- where Ci represents a set of all sensor nodes in a cluster, x represents a value of a coordinate of a sensor node, xi represents a value of a coordinate of a cluster center, k represents the number of the clusters and ranges from 1 to n;
- in a case that a difference of SSEs of two adjacent clusters reaches maximum, determining the number of the clusters as the required number k of clusters, wherein k is calculated by using the following equation:
- k
- =
- argmax
- k
-
- Δ
-
-
- S
-
- (
- k
- )
- -
- Δ
-
-
- S
-
- (
- k
- +
- 1
- )
- where ΔS represents the difference of SSEs of two adjacent clusters, and the number k of the clusters is determined;
- step S12:
- randomly selecting k sensor nodes as initial cluster centers based on the calculated k, calculating SSEs of remaining sensor nodes to the k cluster centers, allocating, for each of the initial cluster centers, sensor nodes from the remaining sensor nodes based on a closest distance rule, and continuously updating the cluster centers and the SSEs until a convergence condition is met, where the cluster centers are updated by using the following equation:
- x
- i
- =
- 1
-
- C
- i
-
-
- ∑
- x
- ∈
- C
- i
-
- x
- and the convergence condition is expressed as:
- |SSE1−SSE2|<ε
- where |Ci| represents the number of sensor nodes in a cluster, ε represents a minimum threshold, SSE1 represents an SSE generated in a current iteration, SSE2 represents an SSE generated in a previous iteration, and coordinates of final cluster centers and sensor nodes included in each of the clusters are determined by continuously performing iteration and updating;
- and
- step S13:
- after iteration, determining, for each of the clusters, a sensor node closest to the cluster center of the cluster as an initial cluster head node of the cluster, and determining the other sensor nodes as ordinary nodes of the cluster;
- and
- wherein a probability of each of sensor nodes in a cluster to be determined as a cluster head node in the next data acquisition period is calculated in step 4 by using the following equation:
- T
-
- (
- s
- )
- =
- {
- 1
- 1
- +
- e
- -
- (
- w
- 1
-
- d
- s
- +
- w
- 2
-
- E
- j
- +
- λ
- )
- if
-
-
- s
- ∈
- G
- ,
- 0
- otherwise
- and the cluster head node in the next data acquisition period is determined based on a calculation result,
- in the above equation, T(s) represents a probability of a sensor node s to be determined as the cluster head node of the cluster in the next data acquisition period, G represents a set of all ordinary nodes in the cluster in a current round, w1 and w2 represent weight constant coefficients, ds represents a reciprocal of a value of a distance between the sensor node s and a center of a cluster corresponding to the sensor node s, Ej represents a ratio of remaining energy to initial energy of the sensor node s, and λ represents an arbitrary constant.