Date: Nov. 28, 1995
Problems: multipath dispersion, security.
3.2 Infrared Links
Optical links can be configured in different ways, depending on
orientation, and beam angle of the transmitter and
receiver. Figure 1 shows the possible configurations. In the cases
(a),(c) and (e) the transmitter (T) and receiver (R) are in
transmitsline-of-sight. The beam can travel directly from the transmitter to
the receiver, without reflection. In the cases (b),(d) and (f) there
is no direct path and before reaching the receiver, the signal is
reflected by the ceiling or ceiling and walls (diffuse reflection).
The directed, line-of-sight configuration is not capable to support
one-to-many and many-to-one connections and it is very unpractical to
adjust transmitter and receiver before transmitting data. The
adjustment of transmitter and receiver could be no major drawback in
the following scenario. In a cell are a fixed number of preadjusted
docking stations incorporating one transmitter and one
receiver. Mobile computers with conventional network cards can be
attached to the docking stations. The docking stations duty is to make
the transformation from the signals of the network card to infrared
signals.
Source [3]
The most challenging configuration but also the one which offers the
most freedom, is the non-directed non-light-of-sight. The transmitter
sends signals in a wide angle to the ceiling and after one or several
reflections the signal arrives at the receiver. The biggest problem in
this configuration is the multipath dispersion: after several
reflections on different paths a short emitted signal is received as a
wide signal. This can cause inter-symbol-interference (ISI) at higher
data rates or larger cells. In this configuration the data rate
depends on the room size. A mathematical model for this effect is
described in [1]. The channel bandwidth-room length product for a room
with 3m-high (9ft) walls is about 60Mbps.m . That limits the data rate in a
5x5m (15x15ft) large room to 12Mbps (60Mbps.m/5m).
Larger bit rates can be obtained with other configurations where the
transmitter sends the signal to a designated area on the ceiling and
the receiver is facing that area (quasi diffuse transmission
figure 1(d)) or where a line-of-sight exists between transmitter and
receiver. In [2] the authors refers to a potential application of
a high-speed wireless LAN figure 2. The Network consists of several
base stations forming a backbone which is connected to a server and
eventually to a wired network, and an arbitrary number of mobile
stations. The base stations are fixed on the ceiling and in
line-of-sight to the mobile stations. In large rooms it is possible to
use more than one base station to cover the whole area. In rooms
without base stations the portable stations should be able to
communicate with one another (ad hoc networking) through a non-directed
non-line-of-sight link, where the ceiling acts as main reflector. To
provide a full duplex connection, different wavelengths or subcarrier
could be used. The latter is better suited for ad hoc networking, as
easier to implement in a portable station.
Source [3]
3.3 Modulation techniques for infrared channels
The following modulation techniques use an intensity modulated
channel. The main difference between an intensity modulated infrared
channel and a conventional wired channel is that the i.m.-channel can
only deal with positive input, and the amplitude of the input signal
is limited not the power.
Different modulation techniques are presented, comparing their power-
and bandwidth-efficiency at a given bit rate.
3.3.1 On Off Keying (OOK)
With OOK, the beam is on when transmitting 1 and off when
transmitting 0. The signal must be coded for being able to distinguish
between the transmission of zeros and the transmitter off status, and
for synchronization.
If the average power is P, then the power used to transmit a 1, must
be 2P.
The bandwidth efficiency is equal to the bit rate (Rb), as one bit can
be transmitted with each signal change.
3.3.2 Two Pulse Position Modulation (2-PPM)
This modulation is identical to the Manchester signaling offset by a
d.c. 1 is represented by the transmission of high-low (here 2,0 and
not 1,-1 because of the offset) and o is coded as low-high.
The power efficiency is the same as with OOK, but the used bandwidth
is doubled. For each bit two signal steps are necessary.
3.3.3 Binary Phase Shift Keying (BPSK)
BPSK transmits a signal in cosine form where 1 is represented by a
phase jump of 180 degrees and 0 is represented by no phase jump. John
R. Barry shows in [3] pp.113-115, that BPSK suffers a power penalty
compared to OOK of 1.5dB due to the additional d.c. and has a
bandwidth efficiency of 2Rb. This power penalty is typical for
intensity modulated channels and not found in conventional wired
environments.
3.3.4 L-level Pulse-Amplitude Modulation (L-PAM)
L different amplitude values of a cosine signal can be transmitted
over a channel. It is possible to code log2(L) bits in each signal
step. For example with L=4 amplitude values 2 Bits can be coded in one
step. As result the used bandwidth is Rb/log2(L).
3.3.5 N-subcarrier Binary Phase Shift Keying (N-BPSK)
When only one subcarrier is used (N=1), then N-BPSK is equal to BPSK.
For N>1 this modulation scheme uses BPSK for each subcarrier.
3.3.6 N-subcarrier Quadriphase-Amplitude Modulation (N-QPSK) or (N-4-QAM)
For every subcarrier 4 different phase jumps are defined.
3.3.7 N-subcarrier L-level Quadriphase-Amplitude Modulation (N-L-QAM)
For every subcarrier L different phase jumps are defined.
3.3.8 L-level Pulse Position Modulation (L-PPM)
This technic has a better power efficiency than the techniques
mentioned earlier. But the needed bandwidth is larger. This method is
widely used in satellite- and fiber-optic-systems.
The input bits are first grouped in blocks of the length log2(L). Then
every block is transmitted by choosing one of L different signal values.
Source [3]
Table 1 shows the power efficiency and bandwidth requirement
for the modulations discussed above.
When we compare the subcarrier- with the baseband-modulations, we
notice that the later have a better power- and bandwidth
efficiency. But the subcarrier modulations have some advantages over
the baseband schemes. By assigning different subcarrier frequencies to
different users, it is possible to achieve asynchronous multiple
access. And subcarrier systems are more robust against multipath
dispersion, since the boud rate in each sub-band is lower.
3.4 Cellular architecture in infrared WLAN's
The topics discussed in the chapter above are related to layer 1 of
the ISO/OSI reference model. In this chapter we will address issues of
the 2nd layer, eg. media access and topologies.
3.4.1 Configuration of one cell
A cell consists of one base station and a number of portable
units. The base station is reachable form every portable unit (in
line-of-sight or not). Conventional topologies can be used for a cell
architecture, eg. bus structures or star configuration.
When the portable units are placed in a way they can 'see' each other,
then normal CSMA/CD could be easily used. If some portable units can
not communicate with others, then it is necessary that the base
station acts as a repeater and sends all received signals back to all
portable units figure 3(a). To avoid interference of the up and down link,
different wavelength or subcarrier could be used
figure 3(b). Wavelength duplex is more complex and expensive to be
used in portable units than subcarrier duplex.
Source [3]
3.4.2 Using more cells
A single cell could be extended to cover the entire network. In this
case an number of base stations cover the network area (eg. a building)
and synchronous the same signals. This technic is called
unison broadcast. This method can only be used for a small number of
portable units and low speeds. Another problem is the multipath
dispersion that occurs when a portable unit receives delayed signals
from two or more base stations.
The main problem in using different cells is that a portable unit could
pass a boundary from one cell to another. We have to assure that the
data exchange is not interrupted, and the portable unit can work in
the new cell. With this aim in view, cells cold be configured in a
way, that they overlap. In two adjacent cells different frequencies
could be used as subcarrier. In the boundary area, a portable unit
receives and sends signals to both cells, so that no dead zone exists.
4 Comparison of Radio Frequency and Infrared Communication
Table 2 summarizes some properties of RF and IR channels.
RF WLAN's are better suited for use in areas where the distance
between transmitter and receiver is big and no high speed
communication is necessary.
IR can be used in areas where radio frequency noise is present
eg. factories or where cell architectures are easy to design like in
office buildings where each cell could be a small room. Security is a
problem in both RF and IR systems and encryption should be used.
IR communication can provide high data speeds because of the high
bandwidth that can be used.
References:
[1] John R. Barry, J. M. Kahn, E. A. Lee, and
D. G. Messerschmitt. 'Simulation of Multipath Impulse Response for
Indoor Diffuse Optical Channels'. Proc. of IEEE Workshop on Wireless
Local Area Networks. May, 1991. Worcester, Ma. USA. pp.81-89.
[2] John R. Barry, J. M. Kahn, E. A. Lee, and
D. G. Messerschmitt. 'High-Speed Nondirective Optical Communication
for Wireless Networks'. IEEE Networkmagazine. November 1991. pp.44-54.
[3] John R. Barry, 'Wireless Infrared Communications'. Kluwer Academic 1994.