The Camera That Fragmented Itself
A video camera streams over 5 GHz to an AP; the link is 80 MHz, 2x2, MCS 8, and the camera RSSI is -58 dBm. A monitor capture shows protected QoS Data frames from the camera (To DS=1, From DS=0), TID 5. The Sequence Control values are 2047/0, 2047/1, 2047/2, then 2048/0. The Retry bit is set on 2047/1 and 2047/2; More Fragments is set on 2047/0 and 2047/1. Frame lengths are about 800 bytes each. Throughput is 18 Mbps, far below the expected rate, and the camera has no wired port. The AP's fragmentation threshold is 2346, but the camera's own configuration is not documented. No A-MPDU aggregation appears in the capture.
1. Interpret the Sequence Control values and the Retry/More Fragments bits. Which frames are fragments of one MSDU, and which are retransmissions?
Read the worked answer
The Sequence Control field is 16 bits: bits 0–3 are the fragment number and bits 4–15 are the sequence number. Values 2047/0, 2047/1, and 2047/2 share sequence number 2047 and increment the fragment number, so they are three fragments of one MSDU. Value 2048/0 starts the next MSDU. More Fragments set on 2047/0 and 2047/1 means more fragments follow; it is clear on the last fragment. The Retry bit set on 2047/1 and 2047/2 means those specific fragments are retransmissions. A repeated sequence/fragment pair with Retry=1 is a retransmission; an incrementing fragment number with the same sequence number is fragmentation. Because only one copy of each fragment appears in this listing, the original failed transmissions were probably not captured or were corrupted before MPDU creation, and the capture shows the successful retries.
2. The AP's fragmentation threshold is 2346, yet the camera sends about 800-byte fragments. What are plausible causes, and how would you confirm? What is the throughput impact?
Read the worked answer
The AP's threshold does not control the camera's transmitter. Fragmentation is a local STA decision, so the camera may have a lower fragmentation threshold, perhaps 800 or 512 bytes, or a vendor reliability setting that fragments video. Confirm by examining every large MSDU from the camera: if all payloads above roughly 800 bytes are consistently fragmented, the parameter is local to the camera. Check the camera's wireless configuration and compare with another client on the same AP. Throughput suffers because each fragment carries its own MAC header, PLCP preamble, and ACK/aggregation overhead and can be retried independently. At 80 MHz MCS 8, 800-byte fragments are inefficient, and the missing aggregation compounds the loss. Setting the camera's fragmentation threshold to 2346, then verifying that large MSDUs are unfragmented and A-MPDU appears, should restore throughput if RF conditions are otherwise good.
3. If the Retry bit were not preserved by the capture driver, how else could you identify retransmitted fragments from the trace?
Read the worked answer
Look for repeated sequence-number/fragment-number pairs, identical frame lengths, and timing that is inconsistent with normal fragment progression. A fragment that appears twice with the same sequence and fragment number is a retransmission even without the Retry bit. Correlate with ACK/Block Ack behavior: a missing MPDU in a Block Ack bitmap followed by a later copy of the same sequence/fragment pair is strong evidence. Also compare timestamps and inter-frame spacing; retransmissions often appear after a timeout or after a PHY error. If the driver strips the Retry bit but preserves sequence numbers and FCS status, the sequence/fragment analysis remains reliable. The absence of the bit weakens per-frame classification but not the duplicate-pair method.