*3.2. Extrapolation of the Experimental Scenario*

The experiments carried out with 22 sensors were then expanded to 100 sensors and 200 sensors. This is in order to predict the results of the energy model by means of an event-based simulator, programmed in C ++ and previously tested in [10].

Next, it is interesting to analyze how the energy model behaves when the network begins to grow in its number of nodes. Table 7 shows three representative protocols out of the six we have studied in this work. Two predominant types of energies were analyzed based on the scalability of the network, from 22 to 300 nodes. Simulations were run for a full day and for nodes which parameters are detailed in Table 3.


**Table 7.** Energy model scalability for AODV, PEGASIS, and MPH protocols.

Results in Table 7 show the two predominant energies for each network and for each of the three protocols studied. Here, we highlight the utility of the proposed model, which breaks down the total energy into the basic energy types of a node. This information becomes relevant when analyzing the skeleton of what is happening in the node with respect to performance metrics such as overhead, collisions, packet loss, interference, complexity of handling neighbor or routing tables, etc.

Analyzing the data, we observe that, when the network is large (about 300 nodes), the AODV and PEGASIS protocols present predominantly CSMA and TX energies. This may suggest that PEGASIS is not a protocol for networks that are too large because the route chain to reach the destination node becomes complex and packets can be lost. Below the 200 nodes, the three protocols under evaluation present the energies of RX and TX as preponderant. Visualizing in detail the network of 100 nodes, the difference between AODV and MPH is 67 % in favor of MPH, and the difference between MPH and PEGASIS is 58% in favor of MPH for the reception energy. This is an example that shows that MPH has a hierarchical topology and that the neighbor tables of the nodes become more manageable thanks to the periodicity of their update. The AODV protocol has the highest energy expenditure and this may be due to the fact that it has a greater number of control packets and if the routes expire or become obsolete, the nodes must start the entire route request process again. In this case, when the networks increase in size, the proposed energy model is very useful because, according to the characteristics of the network, it can establish weak points of behavior and detect possible gaps in energy loss. For example, if CSMA power is increased too much, this may be showing that the communication channel is continuously busy or that this is combined with multiple packet retransmissions, which implies that the processing time increases and the performance of the device is impaired. Then, the breakdown of energies performed by the model can detect abnormal increases or decreases in the behavior of the nodes, as occurs, for example in the PEGASIS protocol when the network is very large.
