Digital Signal Processing Reference
In-Depth Information
be transmitted, the channel access procedure should be carried out. This channel
access relies on a constant sensing of the channel. The channel needs to be sensed
idle for a well-specified duration before the packet can be transmitted.
First, before each packet transmission, an Interframe Space (IFS) needs to be
considered. Then, to prevent STAs from accessing the channel simultaneously, a
random binary exponential backoff counter is introduced. Two different types of
channel or carrier sensing are standardized. The physical carrier-sense is provided
by the PHY, and depends on the PHY Clear Channel Assessment (CCA) procedure.
Further, a virtual carrier-sensing mechanism is provided by the MAC via a Network
Allocation Vector (NAV) counter. This counter can be set since each frame carries
the
Duration
field in the header. Hence, any correctly received frame can result
in an update of that NAV counter, if the new NAV is larger than the current setting.
This NAV overrides the physical layer carrier sensing.
Wireless communication is sensitive to the so-called hidden node problem. It can
be seen in Fig.
2.2
(a) that STA1 and STA3 can hear STA2, but they cannot hear each
other. As a result, their physical layer carrier sensing and their virtual carrier sensing
mechanisms cannot detect ongoing transmissions from each other. However, since
the interference range is typically at least as large as the transmission (and approx-
imately the carrier sense) range, they can interfere with, hence corrupt, ongoing
transmissions of each other. To prevent this, the collision avoidance mechanism has
been proposed (Figs.
2.2
(b) and
2.2
(c)). It consists of two short control frames that
are transmitted before each data frame transmission. The first short control frame is
referred to as Request To Send (RTS), and is sent by the node that wants to trans-
mit a data frame (Fig.
2.2
(b)). If the destination node received the RTS correctly,
it replies with a Clear To Send (CTS) message to inform the source it can send its
data frame (Fig.
2.2
(c)). In the header, the total duration of the planned transac-
tion is listed. STA3, overhearing the CTS, can then set its NAV and delays channel
access for the total transaction duration. As a result, the hidden node problem is
solved. However, this RTS/CTS exchange results in a considerable additional proto-
col overhead for each data frame to be transmitted. As a result, it is not mandatory,
and is often dropped for short packet transmissions, more specifically for packets
that are smaller than the
RTS Threshold
.
It is clear that the hidden node problem can be solved by using a much lower CCA
threshold, or alternatively, transmitting the packets at a much higher power. The
impact of lowering the sensing threshold or the transmission power on the spatial
reuse is a complex problem and not well understood. It is very difficult to derive
analytical models to predict the performance of a network for a given setting of
those thresholds, since the exact performance depends on a lot of parameters such
as traffic density, exact topology etc. As a result, the optimal setting can only be
determined at run-time when more information about the exact operating conditions
is available. This will be illustrated in Chap. 7.
On top of the DCF, an optional PCF can be used. It enables a poll-and-response
MAC for nearly isochronous
1
service. It can be used in the infrastructure BSS only,
1
Isochronous refers to processes where data must be delivered within certain time constraints.
