7.3 Localization Scheme
7.4.2 Effect of h
The value of h is critical to the performance of the proposed scheme. On the one hand, a large value of h allows an unknown node to reach more sensors for help and thus have a higher chance of resolving its location ambiguity. However, on the other hand, it also leads to a high level of communication overhead. Therefore, care must be taken in configuring the value of h to ensure a high possibility of location discovery at a low cost of communications. Figure 83presents the number of resolved nodes after the location discovery completes in networks with various levels of densities. In these scenarios, 3 nodes are initially configured as anchor nodes. The results show that the number of resolved nodes remains the same in most cases. It slightly increases in the network of 50 and 80 nodes when h increases from 2 to 3. The reason is that nodes multi-hop away may be geographically too far away from the unknown node U to provide any useful Out-of-Range information. The value of h is set to be 2 in the rest of the simulations.
7.4.3 Performance Comparison
The number of resolved sensors after the location discovery is completed is used to measure the effectiveness of the proposed scheme in comparison to the basic multi-lateration scheme. The number of additional nodes located using Out-of-Range information depends on the network connectivity and topology. Figure 84 presents the number of resolved nodes after location discovery with Out-of-Range and without Out-of-Range information in a set of
0 10 20 30 40 50 60 70 1 2 3 4 5 6 7
Number of resolved nodes
Number of hops to send LOCATION_HELP message n=80 n=70 n=60 n=50 n=40 n=30
Figure 83: The effect of h in networks with four reference nodes
0 20 40 60 80 100 120 2 3 4 5 6 7 8 9 10
Number of resolved sensors
Network density ML with Out-0f-Range
ML
scenarios. Each data point in Figure 84 is the average value of 20 runs. In each run, a set of three nearby sensors which are neighboring to at least one other sensor are chosen randomly as reference sensors to start the location discovery process. In addition to the average number of resolved sensors, the 5% confidence interval is also shown in the figure. The number of anchor nodes remains at 3 in the simulations.
When the average node degree is small, no Out-of-Range information can be used due to the lack of connectivity in the network. On the other hand, when the average node degree is large, no Out-of-Range information is needed since sensors can estimate their locations using reference nodes. In the other cases, the proposed scheme can locate more nodes than the basic multi-lateration scheme. The results show that as the sensor network connectivity starts to decrease, the Out-of-Range information can be used to locate more sensors in the network.
It is worth noting that the average number of resolved sensors does not always increase as the density increases in the network. This is due to the fact that nodes are uniformly placed over the entire area. With the uniform placement, the area is divided into a number of cells and nodes are randomly placed within each cell. A slight increase in the number of nodes can result in an additional cell with only few nodes placed in it. Selecting reference sensors from this additional cell could end up with very few sensors being resolved after the location discovery process. Nonetheless, a relative big increase of network density does result in an increase of number of resolved sensors after the location discovery process.
In addition to Figure 84 which shows the absolute number of resolved sensors, we also present the percentage of resolved sensors after location discovery process in networks with different levels of densities in Figure85. The data in Figure85shows that with Out-of-Range information, a significant larger percentage of sensors can be located in the networks with medium range densities.
Next, Figure 86 shows how many anchor nodes are required in order to discover the locations of all sensors in the network when the network connectivity is high. As the figure shows, the number of anchor nodes required to discover all sensors can be reduced using Out- of-Range information in networks with low and medium density. In high density network, there is no need to use Out-of-Range information for location discovery since sensors can
0 0.2 0.4 0.6 0.8 1 1.2 2 3 4 5 6 7 8 9 10
Percentage of resolved sensors
Network density ML with Out-0f-Range
ML
Figure 85: Percentage of resolved sensors after location discovery
2 4 6 8 10 12 14 16 18 2 3 4 5 6 7 8 9 10
Number of refenrece sensors needed
Sensor network density ML ML with Out-of-Range
gain sufficient information about reference sensor nodes and compute their locations using multi-lateration.
7.5 SUMMARY
This chapter presents the location discovery scheme used in the framework for semantic view processing. The location discovery scheme is based on “Out-of-Range” information and it is shown how this information can be used to resolve sensor location ambiguities when combined with multi-lateration. The simulation results show that with out-of-range information, fewer reference nodes are needed to locate sensors in the network, which in turn reduces cost and energy consumption of the whole network since reference nodes are usually much more expensive and consume more energy.
8.0 SECURE MESSAGE EXCHANGE FOR SEMANTIC VIEW