Changes between Version 7 and Version 8 of 802.11/wlan_exp/app_notes/tutorial_token_mac/extensions
- Timestamp:
- Jul 13, 2015, 4:45:29 PM (9 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
802.11/wlan_exp/app_notes/tutorial_token_mac/extensions
v7 v8 247 247 == Characterizing Extended TokenMAC == 248 248 249 So, how does this extended TokenMAC perform? We repeat the experiment from [wiki:802.11/wlan_exp/app_notes/tutorial_token_mac/characterization the characterization section]. 250 249 251 || [[Image(wiki:802.11/wlan_exp/app_notes/tutorial_token_mac/figs:token_xput.png, width=600)]] || [[Image(wiki:802.11/wlan_exp/app_notes/tutorial_token_mac/figs:token_extended_xput.png, width=600)]] || 250 252 || '''TokenMAC''' || '''Extended TokenMAC''' || 251 253 252 === 1. Transition A: Beginning of Time === 254 The above figure shows a considerable improvement in the regimes where only one traffic flow is enabled, achieving over 15 Mbps where our standard TokenMAC implementation achieved only 8 Mbps. To explain why, we use the same packet log visualization as the previous section in the tutorial to view the A, B, C regions annotated in the above results. 255 256 === Transition A: Beginning of Time === 253 257 254 258 || [[Image(wiki:802.11/wlan_exp/app_notes/tutorial_token_mac/figs:token_extended_A.png, width=1200)]] || 255 259 || '''Extended TokenMAC: Transition A Timeline''' || 256 260 257 === 2. Transition B: Start of a Contending Traffic Flow === 261 At the beginning of the experiment, neither that AP nor the STA is transmitting anything. As such, our adaptive protocol has deemed neither device worthy of having the {{{TOKEN_RES_MULT_FACTOR_MAX}}} multiplication factor on its reservation period. When the AP starts transmitting, you can see a relatively small window of transmissions on the far left of the above figure. At the end of this period, the AP performs its performance audit and decides to set the multiplication factor to {{{TOKEN_RES_MULT_FACTOR_MAX}}} on the next reservation period. From that point forward, we can see that the relative duty cycle of the AP's reservation period to the idle period reserved for the STA is very large. Because our protocol gives less time to the STA, we are able to achieve a significant boost in overall throughput since we are not wasting nearly as much time as the basic TokenMAC implementation. 262 263 === Transition B: Start of a Contending Traffic Flow === 258 264 259 265 || [[Image(wiki:802.11/wlan_exp/app_notes/tutorial_token_mac/figs:token_extended_B.png, width=1200)]] || 260 266 || '''Extended TokenMAC: Transition B Timeline''' || 261 267 262 === 3. Transition C: Removal of a Contending Traffic Flow === 268 When the STA begins transmitting, we see a small burst of transmissions during one of its reservation periods. At the end of this period, the AP recognizes that the STA is now utilizing the time given to it and decides to increase its multiplication factor to {{{TOKEN_RES_MULT_FACTOR_MAX}}} for the next reservation. From that point forward, we can see that both the STA and the AP continue to achieve the maximum multiplication factor and continue to split the medium evenly. 269 270 === Transition C: Removal of a Contending Traffic Flow === 263 271 264 272 || [[Image(wiki:802.11/wlan_exp/app_notes/tutorial_token_mac/figs:token_extended_C.png, width=1200)]] || 265 273 || '''Extended TokenMAC: Transition C Timeline''' || 274 275 When the STA stops transmitting, we see that the final part of its reservation period goes idle. There was still enough receptions during this time for the AP to justify continuing to give the STA a {{{TOKEN_RES_MULT_FACTOR_MAX}}} multiplication factor for the next reservation. When this reservation period goes unused, the AP drops the STA's multiplication factor back to {{{TOKEN_RES_MULT_FACTOR_MIN}}}, thereby increasing the relative duty cycle of the AP's own transmissions. 276