Link Adaptation Scheme

3.5 Link Adaptation Scheme

Note that the higher the PHY mode that are used (m↑), the shorter the transmission time will be, but the less likely the delivery will succeed within the frame retry limit ( ↓). So, for any given wireless channel condition, there exists a corresponding PHY mode that maximizes the expected effective goodput.

Such a PHY mode is called the best transmission strategy for data frame delivery under the given wireless channel condition. Therefore, an LA algorithm for improving the goodput performance of the IEEE 802.11a WLAN is needed. Its principle is to improve the efficiency of a system by adapting the modulation scheme to the current link condition.

Using the IEEE 802.11a PHY parameters listed in Table 2.2 in the equations derived above, Figure 3.7(a) and (b) show the numerical results of the effective goodput, according to the goodput analysis of different PHY modes with the MSDU size of 200 octets and 2,000 octets in the network consisting of five contending stations, respectively. As expected, the higher rate PHY mode results in better goodput performance in the high SNR range, while the lower rate PHY mode results in better

goodput performance in the low SNR range. One interesting observation is that the PHY mode 3 (QPSK modulation with rate-1/2 coding) always achieves better effective goodput than PHY mode 2 (BPSK modulation with rate-3/4 coding) under all SNR conditions for both frame sizes. The reason for this is that, although QPSK has a higher BER compared to BPSK, the better performance of the rate-1/2 convolutional code compared to rate-3/4 convolutional code compensates it (see Figure 3.5). Therefore, in the presence of the PHY mode 3, the PHY mode 2 may mot be a good choice for data delivery services. Another observation from Figure 3.7 is that a smaller MSDU size results in lower effective goodput due to the fixed amount of MAC/PHY layer overheads for each transmission attempt. Figure 3.8 shows the maximum effective goodput and the corresponding PHY mode selections for different SNR values. Notice that the PHY mode 2 is not part of the selections, which is consistent with the fact that the PHY mode 2 results in a smaller effective goodput than PHY mode 3 under all SNR conditions, as shown in Figure 3.7. In Figure 3.8, we can clearly observe the operating range for each PHY mode. Using these observations in Figure 3.5, the BERs for each PHY mode in their own operating range are quite small (<10^-4) compared to the collision probability (>10^-1, see Figure 3.3), which confirms the assumption in Section 3.1 that we ignore the effect of bit errors introduced by channel noise to derive the collision probability. This also implies that the operating range for each PHY mode ensures the link reliability with lower BER (or packet error rate).

Furthermore, we can use the same analytical method on the basic access method and derive the similar result, Figure 3.9, as Figure 3.8. Comparing these two figures, we observe that in the low SNR range, the RTS/CTS access method outperforms the basic access method with 2,000-octet MSDU, but loses its advantage in the higher SNR range. That implies we should extend our link adaptation algorithm to select MAC mechanisms as well. Table 3.1 and 3.2, derived from the comparison of some

numerical results with different MSDU size, MAC mechanisms and contending stations, list the selected PHY mode with the proper MAC mechanism corresponding to the given SNR range. From these two tables, an evident conclusion can be made.

The longer the data payload is transmitted, the more likely the RTS/CTS access method is used. It happens to agree completely with the idea of using RTS_Threshold.

But another observation makes the RTS_Threshold, the only parameter to decide whether the RTS/CTS access method is applied, insufficient. The observation is that even a 2,000-octet MSDU is transmitted (the upper limit of payload size is 2304 octets). The use of the basic access method is preferable in the high SNR range with the higher PHY mode. On the other hand, a 200-octet MSDU may be transmitted in the low SNR range will use the RTS/CTS access method with lower PHY mode. The rational behind this is that, the transmission duration of the data frame is the key factor not the length. Comparing Table 3.2 with 3.1, the more contending stations in the network, the more likely the RTS/CTS access method is chose because of the more potential collision events. Figure 3.10 shows the effect of the number of contending stations. The effective goodputs with different PHY modes decrease as the contending stations increase.

Figure 3.11 shows the system architecture for adopting LA algorithm. The link adaptor provides three levels of functionality. First, the link adaptor estimates the current SNR condition and the number of contending stations by monitoring the channel via the routing broadcast frames and previous transmission results. Second, the link adaptor selects the optimal combination of the PHY mode and MAC mechanism based on the SNR estimation, length of the data frame to be transmitted, and number of contending stations (see Table 3.1 and 3.2). The three functionalities are represented in the figure as the channel information estimator, MAC mechanism selector, and PHY mode selector, respectively.

Notice that this architecture is transparent to higher layers and can typically be implemented in existing networks. This makes it compatible with existing networks or higher layer applications. Besides, the basic idea of LA is to take advantage of the different modulation schemes, and FEC capabilities provided by IEEE 802.11a PHY and different MAC mechanisms provided by IEEE 802.11 MAC. Therefore, the implementation of the link adaptor should be fairly simple, which makes this algorithm more appealing.