Why are the results not stable?
Code depends on delay differences between packets
Random IPDV should be small
Train of packets
- All have to be sent out back to back
- Send and received processes run in user space
- Vulnerable to CPU scheduling
- If this happens once during a measurement, then the result is already affected