<<

. 20
( 87 .)



>>



0
#1 #2 #3




RTS/CTS

140


120


100
Throughput [Kbps]




80
UDP1
UDP2
60


40


20


0
#1 #2 #3

Figure 3.15. Type 4 experiments with CBR/UDP traffic.
92 IEEE 802.11 AD HOC NETWORKS: PROTOCOLS, PERFORMANCE, AND OPEN ISSUES


3.5 IEEE 802.11b

The results presented in the previous section have been obtained by considering IEEE
802.11-based ad hoc networks. Currently, however, the Wi-Fi network interfaces are be-
coming more and more popular. Wi-Fi cards implement the IEEE 802.11b standard. It is
therefore important to extend the previous experimental analysis to IEEE 802.11b ad hoc
networks.
The 802.11b standard extends the 802.11 standard by introducing a higher-speed Phys-
ical Layer in the 2.4 GHz frequency band while still guaranteeing the interoperability with
802.11 cards. Specifically, 802.11b enables transmissions at 5.5 Mbps and 11 Mbps, in
addition to 1 Mbps and 2 Mbps. 802.11b cards may implement a dynamic rate switching
with the objective of improving performance. To ensure coexistence and interoperability
among multirate-capable stations, and with 802.11 cards, the standard defines a set of
rules that must be followed by all stations in a WLAN. Specifically, for each WLAN is de-
fined a basic rate set that contains the data transfer rates that all stations within the
WLAN must be capable of using to receive and transmit.
To support the proper operation of a WLAN, all stations must be able to detect control
frames. Hence, RTS, CTS, and ACK frames must be transmitted at a rate included in the
basic rate set. In addition, frames with multicast or broadcast destination addresses must
be transmitted at a rate belonging to the basic rate set. These differences in the rates used
for transmitting (unicast) data and control frames has a big impact on the system behavior,
as clearly pointed out in [7].
Actually, since 802.11 cards transmit at a constant power, lowering the transmission
rate permits the packaging of more energy per symbol, and this increases the transmission
range. In the next subsections we investigate, by means of experimental measurements, (i)
the relationship between the transmission rate of the wireless network interface card
(NIC) and the maximum bandwidth utilization, and (ii) the relationship between the trans-
mission range and the transmission rate.

3.5.1 Available Bandwidth
In this section, we will show that only a fraction of the 11 Mbps nominal bandwidth of
IEEE 802.11b cards can be used for data transmission. To this end, we need to carefully
analyze the overheads associated with the transmission of each packet (see Figure 3.16).
Specifically, each stream of m bytes generated by a legacy Internet application is encapsu-
lated by the TCP/UDP and IP protocols that add their own headers before delivering the
resulting IP datagram to the MAC layer for the transmission over the wireless medium.
Each MAC data frame is made up of (i) a MAC header, say MAChdr, containing MAC ad-
dresses and control information8 and (ii) a variable-length data payload, containing the
upper-layer data information. Finally, to support the physical procedures of transmission
(carrier sense and reception), a physical layer preamble (PLCP preamble) and a physical
layer header (PLCP header) have to be added to both data and control frames. Hereafter,
we will refer to the sum of the PLCP preamble and PLCP header as PHYhdr.
It is worth noting that these different headers and data fields are transmitted at different
data rates to ensure the interoperability between 802.11 and 802.11b cards. Specifically,

8
Without any loss of generality, we have considered the frame error sequence (FCS) for error detection as be-
longing to the MAC header.
93
3.5 IEEE 802.11b


TPayload



Application
m Bytes Layer

Transport
Hdr TCP/UDP payload
Layer

Network
Hdr IP payload Layer

MAC
Hdr+FCS MAC payload
Layer

Physical
PHY Hdr PSDU
Layer



TDATA

Figure 3.16. Encapsulation overheads.



the standard defines two different formats for the PLCP: Long PLCP and Short PLCP.
Hereafter, we assume a Long PLCP that includes a 144 bit preamble and a 48 bit header,
both transmitted at 1 Mbps, whereas the MAChdr and the MACpayload can be transmitted at
one of the NIC data rates: 1, 2, 5.5, and 11 Mbps. In particular, control frames (RTS, CTS,
and ACK) can be transmitted at 1 or 2 Mbps, whereas the data frame can be transmitted at
any of the NIC data rates.
By taking into consideration the above quantities, Equation 3.1 defines the maximum
expected throughput for a single active session (i.e., only a sender“receiver couple is ac-
tive) when the basic access scheme (i.e., DCF without RTS/CTS) is used. Specifically,
Equation 3.1 is the ratio between the time required to transmit the user data and the over-
all time the channel is busy due to this transmission:

m
ThnoRTS/CTS = (3.1)
CW min
DIFS + TDATA + SIFS + TACK + · Slot_Time
2

where
TDATA is the time required to transmit a MAC data frame; this includes the PHYhdr,
MAChdr, MACpayload, and FCS bits for error detection.
TACK is the time required to transmit a MAC ACK frame; this includes the PHYhdr and
MAChdr.
CW min
· Slot_Time is the average backoff time.
2
94 IEEE 802.11 AD HOC NETWORKS: PROTOCOLS, PERFORMANCE, AND OPEN ISSUES


When the RTS/CTS mechanism is used, the overheads associated with the transmission
of the RTS and CTS frames must be added to the denominator of Equation 3.1. Hence, in
this case, the maximum throughput, ThRTS/CTS, is defined as

m
ThRTS/CTS = CW min
DIFS + TRTS + TCTS + TDATA + TACK + 3 · SIFS + · Slot_Time
2
(3.2)

where TRTS and TCTS indicate the time required to transmit the RTS and CTS frames, re-
spectively.
The numerical results presented below depend on the specific setting of the IEEE
802.11b protocol parameters. Table 3.4 gives the values for the protocol parameters used
hereafter.
In Table 3.5, we report the expected throughputs (with and without the RTS/CTS
mechanism) by assuming that the NIC is transmitting at a constant data rate equal to 1, 2,
5.5., or 11 Mbps. These results are computed by applying Equations 3.1 and 3.2, and as-
suming a data packet size at the application level equal to m = 512 and m = 1024 bytes.
As shown in Table 3.5, only a small percentage of the 11 Mbps nominal bandwidth can
be really used for data transmission. This percentage increases with the payload size.
However, even with a large packet size (e.g., m = 1024 bytes) the bandwidth utilization is
lower than 44%.
The above theoretical analysis has been complemented with the measurements of the
actual throughput achieved at the application level. Specifically, we have considered CBR
applications that exploit UDP as the transport protocol. Applications operate in asymptot-
ic conditions (i.e., they always have packets ready for transmission) with constant-size
packets of 512 bytes.
In Figure 3.17, the results obtained from this experimental analysis are compared with



Table 3.4. Value of the IEEE 802.11b Parameters
Slot_Time PHYhdr MAChdr FCS Bit Rate(Mbps)
20 sec 1 sec 192 bits 240 bits 32 bits
1, 2, 5.5, 11
(2.56 tslot) (2.4 tslot) (0.32 tslot)
DIFS SIFS ACK CSWMIN CSWMAX
50 sec 10 sec 112 bits + PHYhdr 32 tslot 1024 tslot




Table 3.5. Maximum Throughput at Different Data Rates
m = 512 Bytes m = 1024 Bytes
No RTS/CTS RTS/CTS No RTS/CTS RTS/CTS
11 Mbps 3.337 Mbps 2.739 Mbps 5.120 Mbps 4.386 Mbps
5.5 Mbps 2.490 Mbps 2.141 Mbps 3.428 Mbps 3.082 Mbps
2 Mbps 1.319 Mbps 1.214 Mbps 1.589 Mbps 1.511 Mbps
1 Mbps 0.758 Mbps 0.738 Mbps 0.862 Mbps 0.839 Mbps
95
3.5 IEEE 802.11b


11 Mbps UDP
3.5


3


2.5
Throughput [Mbps]


ideal
2
real UDP

1.5


1


0.5


0
no RTS/CTS RTS/CTS

Figure 3.17. Comparison between the theoretical and the measured throughput.



the maximum expected throughputs calculated according to Equations 3.1 and 3.2. The
real throughput is very close to the maximum throughput computed analytically. Similar
results have been obtained by comparing the maximum throughput according to Equa-
tions 3.1 and 3.2 when the data rate is 1, 2, or 5.5 Mbps, and the real throughputs mea-
sured when the NIC bit rate is set accordingly.

3.5.2 Transmission Ranges
The dependency between the data rate and the transmission range was investigated by
measuring the packet loss rate experienced by two communicating stations whose net-
work interfaces transmit at a constant (preset) data rate. Specifically, four sets of measure-
ments were performed corresponding to the different data rates: 1, 2, 5.5, and 11 Mbps. In
each set of experiments, the packet loss rate was recorded as a function of the distance be-
tween the communicating stations. The resulting curves are presented in Figure 3.18.
Figure 3.19 shows the transmission-range curves derived on two different days (the
data rate is equal to 1 Mbps). This graph highlights the variability of the transmission
range depending on the weather conditions.
The results presented in Figure 3.18 are summarized in Table 3.6, where the estimates
of the transmission ranges at different data rates are reported. These estimates point out
that, when using the highest bit rate for data transmission, there is a significant difference
in the transmission range of control and data frames. For example, assuming that the
RTS/CTS mechanism is active, if a station transmits a frame at 11 Mbps to another station
within its transmission range (i.e., less then 30 m apart) it reserves the channel for a radius
of approximately 90 (120) m around itself. The RTS frame is transmitted at 2 Mbps (or 1
Mbps), and, hence, it is correctly received by all stations within the transmitting station™s
range, that is, 90 (120) meters.
96 IEEE 802.11 AD HOC NETWORKS: PROTOCOLS, PERFORMANCE, AND OPEN ISSUES




Figure 3.18. Packet loss rate as a function of the distance between communicating stations for dif-
ferent data rates.

<<

. 20
( 87 .)



>>