PHYSICAL
LAYER
Now
let’s look at what happens in the physical layer inside
the modem and transceiver IC. The physical layer takes
care of encoding bits to send and decoding received
bits with a base-band modem and radio transceiver.
In fact, this isn’t the entire truth. Things tend
to get messy when you go to implement a conceptual
model in an efficient manner. There are some raw resources
included on the transceiver chip that are actually
part of the MAC layer. These hardware resources off-load
some of the work that otherwise must be performed
on the microcontroller in software.
Other
facilities on the transceiver IC have to do with information
obtained at the physical layer but used at the MAC
layer. For example, received signal strength indication
(RSSI) is used for link quality indication (LQI) to
control power settings. The clear channel assessment
signal is used to implement CSMA-CA functionality.
Now
let’s set aside these extraneous features and get
on with the core job of the transceiver. A DSSS modulator,
in which groups of bits are represented by a symbol,
generates the modulation of the raw data bits. The
symbols are translated into a higher number of bits
by mapping them through a look-up table of larger
bit-sequences chosen for their mathematical properties.
The desired properties include short-run DC balance,
autocorrelation, cross-correlation properties, and
enough apparent randomness to make the waveform appear
as flat noise to a receiver that isn’t supposed to
be listening.
The
reason for discreteness is that a nearby network needs
to ignore the signal to concentrate on the transmissions
from its own network. In systems where the chipping
table constantly changes on a pseudo-random basis,
on-air security is also a prime motivator.
In
the 802.15.4 standard, the raw data bits are grouped
by nibbles to represent symbols. Because 4 bits are
represented at a time, there are 16 different symbols
in the look-up table numbered from zero to 15. Each
symbol corresponds to a 32-bit sequence called a chipping
code. Figure 5 illustrates this process using the
chipping code for the zero symbol.
|
(Click
here to enlarge)
|
Figure
5—The “O” in OQPSK means the I and Q channels
are offset by half a chip period. This limits
the possible instantaneous phase-shifting to 90°
(as opposed 180° with straight QPSK).[2] This
provides a more constant RF envelope and eases
implementation of the power amplifier. |
Each
symbol now consists of a chipping code of 32 bits
called chips, and the rate at which the signal changes
has increased greatly, which spreads the signal over
a wider bandwidth. After some filtering to reduce
the bandwidth, the chipping codes are presented to
the modulator, which carries out half-sine pulse construction.
Offset
quadrature phase-shift keying (OQPSK) is used for
the 2.4-GHz physical layer. There are two sine-based
carriers used in OPQSK. One is in-phase (I) and the
other is in-quadrature (Q), which means it’s offset
by 90°. So, there are sine-based and cosine-based
components with which to represent a symbol. This
is advantageous because the chipping code can be split
and the two halves can be sent simultaneously. The
even chips are represented by the I component and
the odd chips by the Q component. The I and Q waveforms
are added together and amplified before they’re sent
through the transmit/receive switch to the antenna.
Data
represented by multiple bytes is presented least significant
byte first, except for fields associated with security,
in which case it’s the other way around. The entire
process is reversed at the receiver, which is chip-synchronized
with the transmitter and attempts to match one of
the 16 possible codes to values in the datastream.
The closest fitting chip sequence is selected using
a statistics-based maximum likelihood technique. This
results in the dispreading of the correlated signal
in the frequency domain, and the dispersing of any
single narrow band interference. This processing gain
represents a mathematically powered improvement in
the signal to noise ratio.
Figure
6 (see page 22) shows how the spreading and dispreading
rendered by modulation/ demodulation minimizes the
undesirable effects. The horizontal axis is the frequency.
|
(Click
here to enlarge)
|
Figure
6—DSSS demodulation despreading is the
true geek’s version of an amplifier because the
signal has been effectively amplified above the
noise by brains instead of brawn. |
The
chipping rate for the 2.4-GHz PHY is 2 million chips
per second. Because 32 chips are sent for every 4
bits of real data, the effective data rate is as follows:
For
the 868- and 915-MHz physical layers, the modulation
is a binary phase-shift keying (BPSK) and the chipping
rate is 0.3 million chips per second. BPSK is simpler
because the raw data bits simply alter the instantaneous
phase of the carrier. However, the suffix-b proposal
may introduce O-QPSK modulation to the lower bands.
Figure 7 is a block diagram of a typical application.
