Flow monitor in AODV and OLSR

85 views

Skip to first unread message

Maz

unread,

Jan 27, 2016, 2:40:11 PM1/27/16

to ns-3-users

Hi,

I have been trying to compare throughput of AODV and OLSR. My scenario contains 30 nodes distributed uniformly in 300*1500 area with constant position model.15 flows defined, each sending 10 packets per sec (each packet 64 Bytes). I use flow monitor stats to compare.

Issues / questions:

- Delivery Ratio (rxPackets / txPackets) for AODV is always lower than OLSR, no matter what the parameters are (NetDiameter, Hello Interval, …). I have also tried for higher mobility and pause times, which makes the difference in PDR larger. Shouldn't we expect nearly same PDR for constant position and low flow rate?

- The other thing is that number of txPackets in OLSR is always less than AODV, which shouldn’t be since they use similar applications (on/off). Is it possible that flow monitor doesn’t count some of the txPackets for OLSR (may be in case no route found)? or Counts cached packets (because of no route found) for AODV twice (once for the first try and once after receiving RREP)? As you can see in below results, txPackets for OLSR is different for different flow, while it is almost the same for AODV, which makes the first assumption more probable. But how can I find out or fix?

- Flow Monitor, captures control messages for AODV but not for OLSR. Is there a way to get control messages for OLSR by flow monitor?

Below are sample results for AODV:

flowID1 rxPackets: 1688 txPackets: 1853

flowID16 rxPackets: 1434 txPackets: 1853

flowID21 rxPackets: 1392 txPackets: 1852

flowID33 rxPackets: 744 txPackets: 1852

flowID41 rxPackets: 1810 txPackets: 1851

flowID42 rxPackets: 925 txPackets: 1851

flowID43 rxPackets: 1565 txPackets: 1851

flowID52 rxPackets: 1621 txPackets: 1849

flowID62 rxPackets: 1013 txPackets: 1846

flowID63 rxPackets: 1745 txPackets: 1846

flowID73 rxPackets: 1439 txPackets: 1846

flowID79 rxPackets: 1436 txPackets: 1845

flowID84 rxPackets: 1700 txPackets: 1845

flowID85 rxPackets: 1686 txPackets: 1845

flowID86 rxPackets: 1808 txPackets: 1845

for OLSR:

flowID1 rxPackets: 1674 txPackets: 1763

flowID2 rxPackets: 1820 txPackets: 1852

flowID3 rxPackets: 1837 txPackets: 1851

flowID4 rxPackets: 1626 txPackets: 1684

flowID5 rxPackets: 1684 txPackets: 1804

flowID6 rxPackets: 1657 txPackets: 1802

flowID7 rxPackets: 1460 txPackets: 1556

flowID8 rxPackets: 1393 txPackets: 1471

flowID9 rxPackets: 1838 txPackets: 1845

flowID10 rxPackets: 1693 txPackets: 1761

flowID11 rxPackets: 1715 txPackets: 1760

flowID12 rxPackets: 1468 txPackets: 1619

flowID13 rxPackets: 1533 txPackets: 1615

flowID14 rxPackets: 1485 txPackets: 1620

flowID15 rxPackets: 1030 txPackets: 1213

I appreciate your help.

Thanks

Maz

Tommaso Pecorella

unread,

Jan 27, 2016, 7:39:17 PM1/27/16

to ns-3-users

Hi,

answers in-line.

On Wednesday, January 27, 2016 at 8:40:11 PM UTC+1, Maz wrote:

Hi,

I have been trying to compare throughput of AODV and OLSR. My scenario contains 30 nodes distributed uniformly in 300*1500 area with constant position model.15 flows defined, each sending 10 packets per sec (each packet 64 Bytes). I use flow monitor stats to compare.

Issues / questions:

- Delivery Ratio (rxPackets / txPackets) for AODV is always lower than OLSR, no matter what the parameters are (NetDiameter, Hello Interval, …). I have also tried for higher mobility and pause times, which makes the difference in PDR larger. Shouldn't we expect nearly same PDR for constant position and low flow rate?

Not really. Low flow rate could lead to routes invalidation, and re-establishing a route could lead to packet loss.

Moreover, there are some open bugs that could affect your results (check bugzilla).

- The other thing is that number of txPackets in OLSR is always less than AODV, which shouldn’t be since they use similar applications (on/off). Is it possible that flow monitor doesn’t count some of the txPackets for OLSR (may be in case no route found)? or Counts cached packets (because of no route found) for AODV twice (once for the first try and once after receiving RREP)? As you can see in below results, txPackets for OLSR is different for different flow, while it is almost the same for AODV, which makes the first assumption more probable. But how can I find out or fix?

No idea tbh. But see the next answer.

- Flow Monitor, captures control messages for AODV but not for OLSR. Is there a way to get control messages for OLSR by flow monitor?

Nope, because OLSR control messages are broadcasted, and FlowMonitor does NOT count packets directed to multicast or broadcast addresses. Note that also AODV control messages are only partially counted, as some are broadcasts (and some unicast).