Impact of DCF on Fairness Notions

802.11's DCF (Distributed Coordinating Function) is far-and-away the most commonly used contention resolution method in 802.11 networks. Although an alternative Point Coordinating Function (PCF) exists, it is not implemented by most AP vendors because of its complexity and issues of co-existence with DCF-based networks. DCF gives equal transmission opportunities (or long-term channel access probability) to each contender [17,26].

Therefore, competing nodes attempting to send data packets to the AP over the same time interval will be able to transmit equal numbers of frames. DCF's transmission opportunity based mechanism provides fair allocations of both throughput and channel occupancy time only if all contending nodes i) use the same date rate, ii) use the same packet size, and iii) experience very similar loss characteristics. If only the last two conditions hold, DCF achieves throughput-based fairness but does not achieve time-based fairness. For any other combination, DCF achieves neither time-based fairness nor throughput-based fairness.

Figure 4: UDP and TCP throughputs achieved by three competing nodes (Cisco-350 cards) each of which is exchanging data at 11 Mbps with a common AP (Cabletron Roamabout-2000). ``Up'' and ``Down'' x-axis labels denote that the nodes are sending data to and receiving from the AP respectively.

Figure 4 shows the throughputs achieved by three competing nodes that are either sending or receiving data using the maximum data rate of 11 Mbps and the maximum packet size of 1500 bytes. In uplink directions, the throughput achieved by each node is approximately equal due to DCF. In downlink directions, the throughput achieved by each node is approximately equal largely due to the AP queuing scheme, which usually transmits to wireless clients in a round-robin manner. TCP throughputs are significantly less than UDP throughputs because the transmission overhead of TCP ack packets. The total throughputs achieved in the uplink direction are higher than those in the downlink direction. This is because one 802.11 sending node (the AP) cannot fully utilize or saturate the channel since a transmitting node is required to back-off for a random period, between 0 and 610 us, after every successful packet transmission. This overhead is reduced with the increase in number of competing nodes.

We now derive the general expression of T(i), the fraction of time node i is able to transmit or receive packets under DCF. For ease of notation, we will use $\gamma_i$ in place of $\gamma(d_i,s_i,I)$. For steady state performance, we can assume that in each round, each competing node transfers a single packet. Thus, T(i) is simply the ratio of the time required for node i to transfer a data frame, which is $\frac{s_i}{\gamma_i}$, to the total time required for every node in I to transfer a data frame.

$$T(i) = \frac{\frac{s_i}{\gamma_i}}{\sum_{j \in I}{\frac{s_j}{\gamma_j}}}$$ (4)