OLD | NEW |
(Empty) | |
| 1 @node Introduction |
| 2 @chapter Introduction |
| 3 @anchor{chap:physim-wifi:introduction} |
| 4 |
| 5 This module contains a physical layer implementation of the OFDM-based IEEE 802.
11 standard, more precisely, for the Orthogonal· |
| 6 frequency division multiplexing (OFDM) PHY specification for the 5 GHz band@foot
note{see Section 17 of the IEEE 802.11 (2007) standard}.· |
| 7 It can be used as a drop-in replacement for the official @code{YansWifiPhy} impl
ementation when higher simulation accuracy is required. |
| 8 |
| 9 The NS-3 default physical layer, @code{YansWifiPhy}, implements a packet-level P
HY model which abstracts channel effects on individual packet bits·· |
| 10 by using average bit-error rates w.r.t. signal-to-noise and interference ratios
to determine whether a packet is successfully received. In contrast, |
| 11 the @code{PhySimWifiPhy} implementation performs all signal processing steps th
at a real transceiver would follow when decoding a frame. As such,· |
| 12 individual bits are explicitly considered and detailed lower-layer techniques su
ch as bit interleaving, forward error correction, OFDM modulation and so· |
| 13 on are applied. The end result of accounting for these mechanisms is a detailed
and accurate signal representation, allowing consideration of effects· |
| 14 such as frequency- and time-selective fading as well as enabling evaluation of t
he impact of advanced physical layer signal processing algorithms |
| 15 on the performance of the whole network. Further, by modeling the physical layer
at this granularity, existing and new wireless· |
| 16 channel models can easily be implemented and plugged into the simulator, without
the need to build empirical bit-error or packet-error rates.· |
| 17 For additional information on the motivation of this work, consult the publicati
on |
| 18 @uref{http://ieeexplore.ieee.org/search/srchabstract.jsp?tp=&arnumber=5506341,,"
Bridging the Gap between Physical Layer Emulation and Network Simulation"}· |
| 19 which was presented at the IEEE Wireless Communications and Networking Conferenc
e 2010 in Sydney, Australia. |
| 20 |
| 21 @ref{fig:physim-architecture} shows the conceptual architecture of the @code{Phy
SimWifiPhy} implementation, how it interacts with the existing WiFi MAC· |
| 22 layer implementation and which sub-modules (i.e. signal processing modules) are
used to simulate the frame construction and frame reception of a·· |
| 23 transceiver. In the following, the manual gives a basic overview of the frame tr
ansmission and reception process. Further details are then provided in· |
| 24 @ref{Implementation}.·· |
| 25 |
| 26 @float Figure,fig:physim-architecture |
| 27 @caption{Architecture of the PhySimWifi implementation: how it connects to the e
xisting WiFi MAC implementation and which sub-modules are used to· |
| 28 simulate the frame construction and frame reception process.} |
| 29 @center @image{figures/figure1-architecture, 4.5in} |
| 30 @end float |
| 31 |
| 32 Whenever the MAC layer triggers a @code{SendPacket()} request on a @code{PhySimW
ifiPhy} instance, the frame construction process is started and a transition fro
m· |
| 33 packet-level to bit-level and from bit-level to signal-level is performed. To do
so, the data bits of the packet are taken as an input (if given; if not,· |
| 34 a random bit sequence that corresponds to the payload length is generated) and t
he functionality of several sub-modules is used to achieve the transformation.· |
| 35 The necessary sub-modules are called @code{PhySimConvolutionalEncoder}, @code{Ph
ySimBlockInterleaver}, @code{PhySimScrambler} and @code{PhySimOFDMSymbolCreator}
,· |
| 36 and they correspond to the transformations specified in the IEEE 802.11 standard
for OFDM-based transmission. @ref{Frame Construction Process} of |
| 37 this manual elaborates further on the details of the frame construction process.······ |
| 38 |
| 39 After the complex time samples, which constitute the packet, have been generated
, the packet is passed down to the wireless channel (which is of type· |
| 40 @code{PhySimWifiChannel}). The channel computes the propagation delay (using the
existing @code{PropagationDelayModel} implementations), applies a propagation· |
| 41 loss (using sub-classes of @code{PhySimPropagationLossModel}) and schedules a co
rresponding @code{StartReceive()} event at the receiving @code{PhySimWifiPhy} |
| 42 @footnote{The term `receiving' might be misleading here, since the frame might a
rrive, but not be received, because the physical layer might not be able to· |
| 43 detect the frame or synchronize to it.}. For further details on how a propagatio
n loss model can manipulate the signal, look at @ref{Modeling the Wireless Chann
el Effects}. |
| 44 |
| 45 When the @code{StartReceive()} event expires, the incoming packet is first added
to the @code{PhySimInterferenceHelper} module, which keeps track of all· |
| 46 currently incoming frames. Afterwards, the reception process begins, depending o
n the current transmission and reception state of the receiving· |
| 47 physical layer. During the reception process, two additional sub-modules are use
d, apart from the ones already mentioned in the transmission process, namely· |
| 48 @code{PhySimSignalDetector} and @code{PhySimChannelEstimator}. Further implement
ation details are described in @ref{Frame Reception Process}.· |
| 49 |
| 50 Note that the increase in simulation accuracy is accompanied by an increase in c
omputional effort. Depending on the WiFi mode used, the number of network nodes
simulated, the amount of packets |
| 51 and data bits transmitted and the used channel models, the simulation can be up
to 1000 or 10000 times slower than the default @code{YansWifiPhy}· |
| 52 implementation. In general, simulation requirements decrease with lower amounts
of transmitted data, higher PHY data rates (transmission mode) and simpler chan
nel models. |
| 53 A future version will include optional support for OpenCL, such that the signal
processing parts can be parallelized, e.g. through the usage of recent GPGPUs· |
| 54 with multiple compute units and cores.· |
| 55 |
| 56 @node System Requirements |
| 57 @section System Requirements |
| 58 |
| 59 The following libraries are required to build and use the @code{PhySimWifiPhy} i
mplementation: |
| 60 @itemize |
| 61 @item @uref{http://itpp.sourceforge.net/,,IT++ Signal Processing Library
}, Version 4.0.6 or higher |
| 62 @end itemize |
| 63 · |
OLD | NEW |