|
|
|
Introduction and History |
|
Data in Wireless Cellular Systems |
|
Data in Wireless Local Area Networks |
|
Internet Protocols |
|
Routing and Ad-Hoc Networks |
|
TCP over Wireless Link |
|
Services and Service Discovery |
|
System Support for Mobile Applications |
|
|
|
|
Wireless Spectrum scarce, shared among many
different users with distinct needs |
|
Need either license to operate in specific
frequency band or use unlicensed frequency band |
|
Unlicensed bands: no limit on number of users,
but rules governing “behavior” |
|
Licenses used to be given away basically for
free, but this became controversial, plus governments saw this as easy
source of revenue….. |
|
Need for international standardization: meetings
every 2 years (WARC), many international standards bodies and regulatory
offices involved |
|
|
|
|
|
FCC: Federal Communications Commission |
|
allocates frequencies in less than 10 GHz range |
|
US agency, but relevant to Canada (Industry
Canada keeps spectrum allocation “compatible with that adopted by the
United States”) |
|
allocations determine which bands to use and
whether to operate in a licensed or unlicensed band |
|
unlicensed bands: can be used subject to
operational procedures |
|
licensed bands: high performance, expensive |
|
|
|
|
|
|
Industrial, Scientific, and Medical (ISM): |
|
915 MHz band (902 - 928 MHz, 26 MHz bandwidth) |
|
only available in North America |
|
highly crowded, expected to become even more
crowded |
|
many existing users are non-spread-spectrum
applications |
|
2.4 GHz band (2.4 - 2.4835 GHz, 83.5 MHz
bandwidth) |
|
available worldwide |
|
lightly loaded, but interference from microwave
ovens |
|
5.8 GHz band (5.725 - 5.85 GHz, 125 MHz
bandwidth) |
|
only available in North America |
|
lightly loaded, radar interference |
|
|
|
|
|
|
Personal Communications Services (PCS): |
|
1.9 GHz band (1910 - 1930 MHz, 20 MHz bandwidth) |
|
part of overall PCS band (1850-1990 MHz) |
|
1910-1920: unlicensed asynchronous or
packet-switched applications |
|
1920-1930: unlicensed synchronous or
circuit-switched applications |
|
open band in Europe around 1.9 GHz for digital
enhanced cellular telephone (DECT) |
|
in US, this band currently occupied by other
users with microwave point-to-point links, who will clear the band in a few
years |
|
|
|
|
|
communications channel is valuable resource |
|
problem: how to utilize resource efficiently |
|
compression |
|
sharing the channel with multiple users (maybe) |
|
intended to utilize overloaded channel more
effectively |
|
multiplexing |
|
sharing the channel with multiple users
(definitely) |
|
designed to use underloaded channel more
effectively |
|
|
|
|
|
Frequency Division Multiplexing (FDM) |
|
Frequency Division Duplex (FDD) |
|
Time Division Duplex (TDD) |
|
Time Division Multiplexing (TDM) |
|
Code Division Multiple Access (CDMA) |
|
Frequency Hopping |
|
Direct Sequence |
|
|
|
|
|
|
idea: divide transmission frequency range
into narrower bands (subchannels) |
|
used widely in telephone, microwave, CATV,
satellite |
|
frequency division duplex: use two subchannels
for communication, one for uplink, one for downlink (AMPS) |
|
time division duplex: use same subchannel for
both directions, take turns |
|
|
|
|
|
TDM gives user access to full channel capacity,
but only for limited time periods, rotating channel among all users |
|
pure TDM wastes bandwidth because stations might
not use slots, therefore allocate time slots dynamically |
|
|
|
|
GSM combines FDM and TDM: bandwidth is
subdivided into channels of 200 kHz, shared by up to eight stations,
assigning slots for transmission on demand |
|
|
|
|
|
|
|
use whole frequency band to transmit data,
everyone can transmit simultaneously |
|
separate multiple transmissions appropriately |
|
cocktail party analogy: how do numerous pairs of
people in a room interact |
|
TDM: all people in room center, taking turns
talking |
|
FDM: people spread around room in clusters,
conversations within clusters simultaneously |
|
CDMA: all people in room center, talking at the
same time, but in different languages |
|
|
|
|
originally developed to foil jamming by military
opponent |
|
idea: divide frequency band into subbands,
transmit data in one subband at each instant, but change frequency subband
frequently |
|
frequency changes pseudo-randomly to thwart
attacker (or to reduce collision likelihood with parallel send attempt) |
|
sender and receiver need to agree on sequence of
frequency changes |
|
|
|
|
|
|
|
|
How to deal with collisions? |
|
voice transfer: loss of small amounts of
information not critical, just drop it (but: QoS!) |
|
data transfer: have to ensure that no data is
lost |
|
hopping sequences within one cell orthogonal, so
only interference possible with transmissions in a neighbor cell |
|
transmissions from other cells low in power,
typically treated as noise and filtered out |
|
channel coding and interleaving of data ensure
that receiver can tolerate a certain number of burst errors. In GSM,
interleaving is done over 8 bursts, and approximately two of these 8 bursts
can get wiped out and recovered at receiver |
|
|
|
|
|
subdivide each individual bit into m signals
(chips) and transmit those using full bandwidth |
|
each station is assigned a unique m-bit code or
chip sequence, which is combined with value of data bit to determine
sending sequence (for a bit value of 1, send chip sequence, for a bit value
of 0 send one’s complement) |
|
receiving station reconstructs data bit from m
samples, tolerates certain amount of interference |
|
effect is that b bits/second of information are
spread out over mb chips/second, form of spread spectrum communication,
operational requirement for ISM bands |
|
|
|
|
|
|
|
Why does direct sequence CDMA work? Answer: chip
sequences are orthogonal |
|
example: |
|
use bipolar notation: binary 0 is -1, binary 1
is +1 |
|
let S denote chip sequence for station S, S its
negotiation |
|
orthogonality: inner product of two chip
sequences S and T (S T) is 0: |
|
|
|
|
|
orthogonality requirement: “as many pairs are
the same as are different” in each chip sequence |
|
it also follows that |
|
|
|
|
|
|
|
|
|
|
|
|
in an ideal, noiseless channel, the capacity
(i.e., the number of stations) can be made arbitrarily large |
|
however, having all chip sequences synchronized
in time is impossible |
|
receiver has to synchronize with sender by
having sender send long enough known chip sequence that receiver can lock
into |
|
all other (unsynchronized) transmissions will
appear as random noise, requiring larger chip sequences |
|
implicit assumption so far was that power level
of all senders are same as perceived by receiver |
|
in mobile environment, power of received signal
depends on distance between mobile and base station |
|
requires power management (typically by base
station) |
|
|
|
|
|
FH has potentially less total interference: |
|
ratio of intracell-to-intercell interference is
two to one |
|
no intracell interference in FH case |
|
no power control necessary to ensure all signals
are received at base with equal strength |
|
external jamming potentially handled more
gracefully by FH |
|
DS requires contiguous wide band, FH spectrum
does not have to be contiguous |
|
FH has somewhat more complex radio control |
|
|
|
|
|
TDM: common radio and modem equipment, at a
given carrier frequency, can be shared among N users at base station |
|
TDM can adopt dynamically to user traffic
demands (statistical TDM), compared to FDM |
|
TDM has less stringent power control
requirements than DS CDMA |
|
time slot structure gives time for measurements
of alternative slots, frequencies, etc. in order to support mobile assisted
or mobile controlled handoff (will be discussed when talking about GSM in
more detail) |
|
|
|
|
|
TDM has more complex RF units |
|
time slots become available periodically |
|
power envelope is therefore periodically
pulsating |
|
TDM requires complex timeslot assignment and
management (again, will become somewhat clearer when discussing GSM) |
|
TDM is more susceptible to errors due to
multipath fading than FDM (higher bandwidth) |
|
|
|
Notes