Equation (5.9) neglects the mutual influences of the ACs to each other the shared operation implies on the individual changes of maximum saturation throughput per AC. This will be discussed in the next section, where the results of the analytical approximations are compared to simulation results for a large number of parameter combinations.

5.1.3.3Results and Discussion

The example with the three ACs and four backoff entities per AC as defined in Table 5.2, p. 68, and illustrated in Figure 5.11 and Figure 5.12, is used to evaluate the approximation of the saturation throughput Thrpshare in a shared scenario. The backoff entities of one of the three ACs are assumed to apply a range of EDCF parameters: the EDCF parameters are changed gradually from the higher to the lower priority such that many different parameter combinations are analyzed. The other two ACs apply always the same EDCF parameters: the backoff entities of the two other ACs are assumed to operate according to the legacy and lower priority EDCF parameter setups. Many different combinations of EDCF parameters and relative priorities can be studied under these assumptions. The definitions of the EDCF parameters of the higher, legacy, and lower priority ACs can be found in Table 5.2. A constant frame body size of 512 byte for all ACs is selected here, RTS/CTS is not used. Note that the parameters CWmax and RetryCnt remain constant for all ACs at any time and are not varied.

The following figures show the resulting throughput per AC as a result of simulation and analytical approximation, where a different numbers of backoff entities per AC have been assumed. Figure 5.15 shows the results for the scenarios with four backoff entities per AC, i.e., 12 backoff entities in total. Figure 5.17 shows results for scenarios with 10 backoff entities with variable EDCF parameter setup, 2 legacy priority backoff entities, and 4 lower priority backoff entities. Therefore, 16 backoff entities in total share a common channel in these scenarios.

legacy priority

low

priority

variable

priority

receiving station

Figure 5.14: Scenario. One backoff entity per station. All stations detect each other. If two or more stations transmit at the same time, a collision occurs.

lines w/o markers: analytical approx.; lines with markers: WARP2 sim. results

Figure 5.15: Saturation throughput per AC with 4 backoff entities per AC.

Finally, Figure 5.19 shows results for scenarios with 2 backoff entities with variable EDCF parameter setup, 10 legacy priority backoff entities, and 4 lower priority backoff entities. Hence, 16 backoff entities in total are again assumed here. Figure 5.15 is discussed in the next Section 5.1.3.3.1, Figure 5.17 is discussed in Section 5.1.3.3.2, and Figure 5.19 is discussed in Section 5.1.3.3.3.

5.1.3.3.14 Variable Priority Backoff Entities against 4 Legacy and 4 Low Priority Backoff Entities

Figure 5.15 shows simulation and analytical results for 28 parameter combinations, where the EDCF parameters of one AC (used by 4 of 12 backoff entities) are varied from higher (left hand side in the figure) to legacy priority, and down to the lower priority (right hand side in the figure), according to Table 5.2. The scenario is depicted in Figure 5.14.

The other 8 backoff entities of the other ACs operate with legacy and lower priority. It can be seen that the analytical results approximate for all the priorities the simulated results with a sufficient accuracy. This result indicates that the Markov model can be used to sufficiently approximate the backoff process from which the saturation throughput has been calculated.

It can be observed from the left hand side of Figure 5.15 that the AC with the variable priority observes the largest throughput (shares of capacity) in scenarios with higher priority EDCF parameters (AIFSN=2, CWmin=7, PF=24/16). However, this share decreases with changed EDCF parameters towards legacy priority. If the 4 backoff entities of the AC with the variable EDCF parameters

operate according to the legacy priority, then the observed share of capacity is the same as for the 4 legacy backoff entities. This is indicated by the simulation results, and confirmed by the analytical approximations (center of Figure 5.15, AIFSN=2, CWmin=15, PF=32/16). As expected, when changing the EDCF parameters down to the lower priority, the share of capacity of this AC decreases down towards the share that is observed by the backoff entities of the lower priority AC (right hand side of Figure 5.15, AIFSN=9, CWmin=31, PF=40/16). In parallel, the legacy priority backoff entities observe increased shares in theses scenarios. This is an expected result: in these scenarios, the legacy priority AC is parameterized such that the 4 legacy backoff entities access the channel with highest priority relative to the other 8 backoff entities, because those backoff entities all operate with the lower priority EDCF parameters. This again is confirmed by the simulation results as well as the analytical approximations.

5.1.3.3.210 Variable Priority Backoff Entities against 2 Legacy and 4 Low Priority Backoff Entities

In contrast to the previous scenario, a different number of backoff entities per AC is assumed in the following, as shown in Figure 5.16. Figure 5.17 shows simulation and analytical results for parameter combinations with 10 backoff entities with variable EDCF parameter setup, 2 legacy priority backoff entities, and 4 lower priority backoff entities, thus, 16 backoff entities operate in parallel here. The observed shares are the same as in the previous scenario from the last section, illustrated in Figure 5.15. The main difference to the previous scenario is that now the backoff entities that slowly reduce their priority from parameter combination to parameter combination (i.e., from the left to the right in the figure), keep their maximum share a longer time (for more parameter combinations).

After some more parameter combinations (left to right), an immediate reduction of throughput share suddenly happens (indicated in the center of the figure).

legacy priority

low variable priority

priority

receiving station

Figure 5.16: Scenario. One backoff entity per station. All stations detect each other. If two or more stations transmit at the same time, a collision occurs.

lines w/o markers: analytical approx.; lines with markers: WARP2 sim. results

Figure 5.17: Saturation throughput per AC with 10 backoff entities with varying EDCF parameters, contending with 2 legacy and 4 lower priority backoff entities.

This is an obvious result. The 10 backoff entities are more dominant than the 4 backoff entities of the previous scenario. Note that of course the capacity share per backoff entity in the AC with variable priority is less than before, as 10 instead of 4 backoff entities are operating according to this AC. In contrast, only 2 instead of 4 legacy backoff entities operate in the scenario shown here. Thus, these 2 backoff entities will observe an increased resulting throughput when they operate with higher priority, relative to the other 10+4 backoff entities (towards the right in the figure). As before, the analytical results and the simulation results confirm each other with sufficient accuracy.

5.1.3.3.32 Variable Priority Backoff Entities against 10 Legacy and 4 Low Priority Backoff Entities

Figure 5.19 shows results for a scenario with 2 backoff entities with variable EDCF parameter setup, 10 legacy priority backoff entities, and 4 lower priority backoff entities, as illustrated in Figure 5.18. 16 backoff entities in total are assumed as in the last sections. The results are again obvious. Although the 2 variable priority backoff entities operate with highest priority at the beginning (indicated in the left of the figure), they do not observe a considerable share. However, the share per backoff entity is larger for any of the 2 backoff entities than the share observed by any of the 10 legacy backoff entities. It should be emphasized that the analytical results and the simulation results deviate more from each other with such parameter combinations, although the approximations show qualitatively the same shares.