The restrictions imposed by traditional network technologies were presented and we showed how new ways for thinking about routing have emerged to overcome these. They include insights and parallels made from observing a number of biological, social and epidemic behaviors. A number of proposals, associated to these metaphors, make use of mobility patterns, pheromone levels, user habits and profiles, relationships and other types of stimulus to offer self-organization, load balancing, adaptability and advanced technology dependent routing. This section is going to perform some concrete evaluations to show and determine the impact of some of network and other important parameters and examine their configuration. To achieve this, the reader is invited to review some optimization and evaluation techniques that are very much relevant to the context of routing in future networks,
Percolation
The percolation theory is inspired from the observation that there is a limit value for a physical material to make a transition between two states called by "critical phenomenon". For instance, water (a fluid) has two states: liquid and gas. A bottle of water may transition from the state liquid to gas when submitted to a higher temperature, namely, at 100°C at sea level. Another example is that of a filter where there is a given alpha number of porous in a stone. When the number of porous reaches a threshold, water, then, passes to the other side of this stone. These probabilistic changes of states are defined according to a percolation model that uses a threshold to determine such transitions. Hence, such strategy would help determining which routing parameter values would cause percolation, or successful knowledge sharing in the context of future 4G networks.
Some works set up a static percolation coefficient value in order to improve routing. The spatial gossip is an example of a routing algorithm that used this to select the forwarding node. Other works chose to evaluate the environment to discover when such algorithm percolates. For instance, one could seek the relation between buffer size and the success delivery rate. Otherwise, one could check if there is a limit buffer size that determines success or no delivery of messages. The analogy in this example associates messages to a fluid in a percolation scenario and nodes to the surface. Consequently, when all the messages start going from the source and reach their destination, one says that routing has percolated.
Diffusion and Chemotaxy
Adolph Fick was among the pioneering researchers who studied extensively the diffusion process. He observed that salt movement occurs from high to low concentration in liquids and defined an equation to express the proportionality between the flow and the spatial gradient of diffusion. Other researchers also studied the diffusion observing a spontaneous particles movement from low to high concentration. However, there is a common concept among these equations: they expressed the movement of cell or substance to obtain equilibrium, considering, in general, the position as a variable or both time and position.
Similarly, Chemotaxy is a movement behavior according to the gradient of concentration, but it is not a spontaneous event. Chemotaxy represents the attraction or retraction among cells due to some substance. It is commonly used in biology to analyze the behavior of human cell, virus or bacteria. However, such behavior has been analyzed and shown to also benefit the routing environment. Routing policies could be seen as the substance that modifies spontaneous movement.
Given that some message forwarding is based on a probabilistic mechanism set according to the encounter frequency of nodes (i.e. PRoPHET). We could evaluate the diffusion by modifying node movement in order to verify whether node mobility could be a stimulus to influence this behavior or not. In other words, we could check whether node mobility increases or not the message delivery rate.
Considering that PRoPHET could also be executed in sensor networks, policies are likely to move a node by several spaces in order to increase the encounter frequency and as a result may be used to improve the delivery ratio. Alternatively, one could set fake information altering encounter frequency, the message delivery decreases, because messages are removed from a buffer before actually finding their destination.
Stigmergy
Pierre-Paul Grassê introduced the Stigmergy concept after studying nest building. He observed that there is an indirect communication used by social individuals in order to coordinate their efforts towards some objective. For example, Ants lay down more pheromone when they find food to enable other ants to detect and react to this stimulus. In summary, they indirectly interact and cooperate to feed (or finding a path in the routing analogy). Although it is a comprehensive behavior, there is lack of mathematical models or equations to describe Stigmergy. Typically, the stimulus is not reached by some well established known equation, one may consider a given variable as stimulus to Stigmergy behavior and verify whether only a node with a fake variable can modify the Stigmergy of all individuals of a group and consequently the environment
Given that the decision mechanism of PRoPHET routing evaluates the number of encounters of neighbors, we could setup the encounter frequency for a single given node with fake information and next observe the success ratio. The encounter frequencies are used by such node as a Stigmergy where the nodes collaborate with this information to route the information.
No comments:
Post a Comment