Blog blog blog
Next Steps ...
| Topic | Description | |
|---|---|---|
| 1 | EG on Satellite Networks | Simulation of EG on a real satellite network |
| 2 | EG on Scheduling Sensor Networks | Study of scheduling on sensor networks |
| 3 | Stochastic Behavior Networks | Study of uncertain (stocasthic) behavior of networks. How to model? |
| 4 | Fault tolerant EG | Routing protocol based on EG must be fail prone (i.e. the network represented by the EG is not accurate anymore) |
| 5 | Development of EG based protocol for real use | How to exchange messages to construct and maintain the EG (e.g. an OLSR |
| 6 | Fastest Journey | Implementation and Evaluation of fastest journey on NS2 |
| 7 | Add a deadline parameter in the model | Set the amount of time to look ahead in the algorithm |
NS2 and ARP in mobile ad hoc protocols. To be or not be?
During the implementation of the EG shortest journey routing protocol, the results showed high values of dropped packets by ARP. All the packets were dropped in a burst way, and a lot in the same time.
- At a first glance, a great amount of dropped by ARP packets were due to the generation of packets during the sleep phase of a node. So, when the node wake up, a lot of packets are waiting to be delivered at the same time, and packets are dropped until the ARP are not resolved. This was fixed adding a restriction, only waked-up nodes are allowed to generate traffic.
- But, we still have a lot of packets being dropped by ARP and IFQ in the shortest journey. And now they occour isolated.
The main reason to these drops is the management of ARP in NS2, which is derived from the original BSD Unix ARP buffer with only a single packet while resolution is ongoing, and dropping the other ones (Very good discussion about that can be found here – read section II.C. ).
* For now, I will keep the default behavior of NS2 to the my ongoing simulations (i.e. maintaining the single packet buffer). In the future I will use the Marcel Ortega patch to disable the ARP resolution in MANET protocols to analyse the differences.
