In IEEE 802.11–based wireless LANs, the rapid growth of video
streaming, AR/VR, and IoT traffic frequently drives the access point
(AP) buffer into congestion, resulting in increased delay, packet loss, and
degradation of service quality. Convention...
In IEEE 802.11–based wireless LANs, the rapid growth of video
streaming, AR/VR, and IoT traffic frequently drives the access point
(AP) buffer into congestion, resulting in increased delay, packet loss, and
degradation of service quality. Conventional bit-based communication
transmits all data as raw bit streams, while most existing semantic
communication schemes employ a fixed bandwidth compression ratio
(BCR) and do not adapt to AP buffer dynamics, which limits their
effectiveness under time-varying traffic loads.
This thesis proposes an Traffic Load–based adaptive-BCR semantic
communication scheme that jointly considers the semantic importance of
data and real-time AP buffer state. AP buffer occupancy is continuously
monitored and classified into traffic-load regions, and a suitable BCR in
the range of 1/6 to 1/20 is selected for each region, with a hysteresis
mechanism to prevent frequent parameter switching near the thresholds.
The proposed scheme is evaluated using an ns-3–based discrete-event
simulation of an IEEE 802.11 WLAN under low, medium, high, and
mixed traffic-load scenarios. Performance is assessed in terms of AP
buffer occupancy, packet loss rate, end-to-end delay, and Effective PSNR
that accounts for packet drops. The results show that the proposed
adaptive-BCR semantic communication effectively alleviates AP buffer
saturation, reduces packet loss and delay, and improves Effective PSNR,
especially under medium and high traffic loads.