I'm using RTS/CTS to prevent further collision when hidden terminal nodes are present, and also I'm reserving the channel for clear transmission. Even using that mecanism my results get packet loss. What are the reasons that lead packet loss under such circunstances? I have 2 in mind: congestion in buffers and interference with no orthogonal channels.
Yes, you are write buffer overflow and interference with no orthogonal channels can cause packet loss. In addition to these the packet loss also depends upon the underlaying channel model used for the simulation.
The wireless channel is susceptible to packet loss. If you have used any relaistic model then packet losses are bound to occur due to path loss, external interference, multipath fading and shadowing effects, even if there is no collision.
It could be many things ...
- other interferers that don't follow RTS/CTS (microwave ovens, Bluetooth, wireless phones)
- channel fading (distance too large, walls, reflections)
- antennas, connectors, cables damaged
- problems with AP association (you have to be associated to receive data packets)
Generally IEEE 802.11 should try to retransmit a few times, maybe you can check with Wireshark what is happening on the channel.
RTS/CTS mechanism does not solve all hidden node problem situations, in some circumstances the CTS packet does not reach all potential hidden nodes. Anyway, if you are sure that your channel is clear (no other interferer in the vicinity), you may start thinking about adjacent channel interference, buffer overload, bad channel state, long distance between your transmitter and your receiver, large number of associated devices ....
Are you communicating with a local destination (connected to the same AP)? or with a remote one?
In addition to my previous reply, the RTS/CTS mechanism only blocks the one hop nodes of transmitter and receiver. But, it does not consider the additive interference model. Let i and j be the transmitter and receiver repectively. According to additive interference model if multiple nodes at more than one hop (say two hop) distance from the receiver j are transmitting at the same time then the combined interference at node j could be sufficient to destroy the packet transmitted by node i.
Theoritically, blocking the nodes which are at the k distant from the receiver can always cause collisions due to additive interference caused by nodes situates at more than k hops
Hi,
Have you cheked on sniffer? whether your RTS ack by other stations ?
For testing UDP the best freeware tool I used was IPERF. It reports UDP loss directly, and by clever sizing of UDP packets (-l command) you can observe UDP loss per packet size.
This is the way I discovered unexpected packet loss for small packets in an early WiMAX implementation. Maybe not too elegant but works fine.
You may wish to kill all the processes that use network on your laptop prior to IPERF tests, or simply boot it in a safe mode with network, as all network activities of other processes will compete with IPERF and cause false losses.
- Badges
- Science topic