Motivation

In section 2.6, we described the MAC bandwidth components framework for analysing the packet transmission process in IEEE 802.11 WLAN networks. BWbusy, BWaccess, BWidle and BWfree are four bandwidth components that serve to describe the bandwidth utilization on the medium. The concept of access efficiency connects these four components together [33]. This metric indicates the efficiency (in terms of the bandwidth required) of a station in accessing the channel medium. The larger the access efficiency, the more efficiently the station can access the medium. Because stations with

different access efficiencies will have different available bandwidths, when a station contends with its neighbour station for access, the access efficiency of all the stations needs to be considered while estimating the available bandwidth.

In section 3.1, we defined the available bandwidth as the maximum MAC layer traffic load of a station that can be transmitted on a channel without causing saturation either to itself or to other stations which share the same channel.

Therefore, we define saturation and congestion as follows:

Definition 4.1: Saturation. Saturation occurs when the free bandwidth of a station equals zero, i.e. when .

BWfree

=0 BWfree

Definition 4.2: Congestion. Congestion occurs when the number of packets

arriving at the queue per second is larger than the number of packets transmitted per second and when this condition persists for more than seconds.

Nin

Nout

=10

confirm

T

To avoid a false indication of the congestion status, a small value of should be not selected. On the other hand, a large value of will cause high packet loss if a prolonged period of congestion occur before detection. However, the issue of how to select the optimal value of is beyond the scope of this thesis. During the experiment, we found that 10 seconds represented a good trade-off between avoiding false indication of congestion and minimizing the packet loss.

confirm

According to these definitions 4.1 and 4.2, if a station can win more transmission opportunities than the number of packets it wants to transmit, there are no packets that need to be stored in the transmit queue. On the other hand, if a station cannot win a sufficient number of transmission opportunities to satisfy the packets that are arriving into the transmit queue, the packets which cannot be transmitted will be stored in the transmit queue. The depth of the transmit queue will increase until it reaches its capacity.

This represents a congestion condition as it leads to a large packet delay and

catastrophic packet losses. However, in this thesis, we use these two terms interchangeably to refer the congestion status.

If a channel has only one station, the relationship between the four MAC bandwidth components is

=1 + _idle

busy BW

BW (4-1)

access idle

free BW BW

BW = − (4-2) For a channel with multiple stations, the relationship between the four bandwidth components is more complex. The reason is that the time used to gain access the medium is shared between the contending stations.

Figure 4-1 MAC bandwidth components where two stations contend for the medium

For example, in Figure 4-1, Station A and Station B both want to transmit a packet at the same time t1. However, Station A picks a random number of 8 for its back-off counter’s initial value and Station B picks a random number of 4 for its back-off counter’s initial value. Station B will transmit its packets when its back-off counter has decremented to zero at time t₂. The back-off process of Station A is halted when it finds that the channel

is busy, due to Station B commencing its transmission at time t2. This back-off process of Station A will continue once it senses that the medium is idle again at time t3. Examining the time used to access the medium Access_A and Access_B shows that it is shared between the stations. Based on this observation, the relationship between ,

and of the multiple stations sharing the same channel is:

BWbusy Because the access bandwidth of these two stations is different, the free bandwidth of the two stations is also different. When one of the stations increases its traffic load, the station with smaller free bandwidth will become saturated earlier. When a station wants to estimate the available bandwidth, it needs to take into account the access bandwidth of all the stations in order to avoid congestion across the network. We will introduce the passive available bandwidth estimation technique in the next two sections based on the order in which stations become saturated when the traffic load equals the available bandwidth.

When the throughput of a station is equal to the available bandwidth, there are three possible scenarios to be considered here: (i) the access bandwidth requirement of the monitoring station is greater than the maximum access bandwidth requirement of other stations which use the same channel; (ii) the access bandwidth requirement of the monitoring station is less than the maximum access bandwidth requirement of other stations which use the same channel and (iii) the access bandwidth requirement of the monitoring station equal to the maximum access bandwidth requirement of other stations which use the same channel. Based of this, we will analysis how to calculate the available bandwidth in the next three sections.