MANET driven by SF cab trace
TOC
1. Trace Overview2. Our focus
3. Coordinates conversion
4. SIR epidemic routing
5. Communication model
6. Animation
1. Trace Overview
The trace on the cabs San Francisco Bay Area is publicly available at CRAWDAD.
The dataset contains GPS coordinates of 536 cabs over approximately 30 days.
The following Google map snapshot shows the San Francisco Bay Area.
2. Our focus
We focus on the cabs moving in the downtown area (3.5km x 3.5km).
The downtown area is indicated by the red square in the following Google map snapshot.
3. Coordinates conversion
The Euclidean distance between two cabs are calculated using the Vincenty's formula (error + - 0.5 mm).
Further, in the case that GPS records have a large void in time, we apply interpolation so that a cab moves at a constant speed from a point to another instead of jumping.
4. SIR epidemic routing
Consider a basic stochastic SIR epidemic routing algorithm, where a piece of information starts to broadcast from a source node.
By analogy to the way a disease spreads in a human network, a node in the MANET can be in any of three states S, I, R: A node that has never received the information from the source is in the state of susceptible (S), in which the node can accept incoming transmissions if such opportunity arises. A susceptible node goes into the state of infected and infectious (I) immediately after it has received a copy of the information. The node in state I keeps transmitting the information, i.e. remain infectious, for a certain time period, after which the node enters into the state of recovered and immune (R). A node in state R stops transmitting the information to other nodes and will ignore all future transmissions of the same information from other nodes.
5. Communication model
Two nodes are dircetly connected if their Euclidean distance is smaller than or equal to the radio range r0.
6. Animation
1. The cabs in San Francisco Bay Area | |||
Legend |
|||
If the above video does not play, download it here. | |||
2. The cabs in San Francisco (Zoom in) | |||
Legend |
|||
If the above video does not play, download it here. |