Escolar Documentos
Profissional Documentos
Cultura Documentos
1. INTRODUCTION
Fig. 1.1 shows the single-cell configuration, a simple wireless LAN strategy typical of many environments. It is so named because all the wireless end systems are within range of a single control module.
A backbone wired LAN, such as Ethernet, supports servers, workstations, and one or more bridges or routers to link with other networks. A control module (CM) acts as an interface to a wireless LAN. The module includes either bridge or router functionality to link the wireless LAN to the backbone and some sort of access control logic, such as a polling or token-passing scheme, to regulate access from the end systems. Some of the end systems are stand-alone devices, such as a workstation or a server. Hubs or other user modules (UMs) that control several stations off a wired LAN may also be part of the configuration.
Another common configuration is a multiple-cell wireless LAN, in which a wired LAN connects multiple control modules [1]. Each control module supports wireless end systems within its transmission range. An infrared LAN, for example, limits transmission to a single room, so each room in an office building would need one cell.
In a Nomadic Access configuration, the wireless LAN links a LAN hub and a mobile data terminal equipped with an antenna, such as a laptop or notepad computer. Thus, for example, an employee returning from a trip can transfer data from a personal portable computer to an office server. Nomadic access is also useful in an extended environment such as a campus or a business operating from a cluster of buildings. In both cases, users can move around with their portable computers and access the servers on a wired LAN from various locations.
In an Ad hoc network configuration, a network is set up temporarily to meet some immediate need. It has no centralized server. Thus, in meetings, a group of employees, each with a laptop or palmtop computer, can link their computers in a network that lasts just as long as the meeting.
The IEEE 802.11 standard for wireless LANs is presently the dominant standard for Wireless LANs. It specifies the implementations of the Medium Access Control (MAC) and physical (PHY) layers.
system can be a switch, wired network, or wireless network. The portal integrates the IEEE 802.11 architecture with a traditional wired LAN. The portal logic is implemented in a device, such as a bridge or router, which is part of the wired LAN and attached to the distribution system. These extensions to the Basic Service Set constitute an Extended Service Set as shown in Fig. 1.2, in which a distribution system connects two or more basic service sets. Typically, the distribution system is a wired backbone LAN, but it can be any communications network. The extended service set appears as a single logical LAN to the logical link control (LLC) level. The access point is the logic within a station that provides access to the distribution system by providing services in addition to acting as a station.
Most of the early 802.11 networks used the FHSS scheme, which is simpler. Networks that used the DS-SS scheme were more effective, but all the original 802.11 products had data rates of at most 2 Mbps, which limited their usefulness. In 1999, the IEEE issued the second and third physical layers, IEEE 802.11a and IEEE 802.11b, at roughly the same time. IEEE 802.11a operates in the 5-GHz band at data rates up to 54 Mbps. IEEE 802.11b operates in the 2.4-Ghz band at 5.5 and 11 Mbps. Because 802.11b is easier to implement, it has yielded products first. IEEE 802.11b extends the IEEE 802.11 DS-SS scheme, providing data rates of 5.5 and 11 Mbps through the use of a more complex modulation technique Complementary Code Keying (CCK). Although 802.11b is successful to some degree, the data rate is still too low for applications that need a truly high speed LAN. IEEE 802.11a targets this specific need. Unlike the other 802.11 standards, it specifies the 5-GHz band, and it replaces the spreadspectrum scheme with the faster orthogonal frequency-division multiplexing. OFDM, also called multi-carrier modulation, uses up to 52 carrier signals at different frequencies, sending some of the bits on each channel. Possible data rates are 6, 9, 12, 18, 24, 36, 48, and 54 Mbps.In 2003, the IEEE issued the fourth physical layer std. IEEE 802.11g. IEEE 802.11g PHY operates in 2.4 GHz band and the possible data rates are 1 and 2 Mbps (using DSSS), 5.5 and 11 (using DSSS/CCK), 6, 9, 12, 18, 24, 36, 48, and 54 Mbps (using OFDM).
ECE Department, GITAM University
2: LITERATURE SURVEY
2.1 Evolution of WLAN Standards
The first wireless Ethernet standard, IEEE 802.11, was adopted in 1997 [6]. This standard provided for three physical layer (PHY) specifications including infrared, 1-2 Mbps frequency hopping spread spectrum (FHSS) and 1-2 Mbps direct sequence spread spectrum (DSSS) in the 2.4 GHz ISM band. Because wired Ethernet LANs at the time were capable of speeds up to 10 Mbps and early products were quite pricey, the original 802.11 standard had limited success in the market.
Two years later, the original 802.11 standard evolved along two paths. The 802.11b specification increased data rates well beyond the critical 10 Mbps mark, maintained compatibility with the original 802.11 DSSS standard and incorporated a more efficient coding scheme known as complimentary code keying (CCK) to attain a top-end data rate of 11 Mbps . A second coding scheme, Packet Binary Convolutional Code (PBCC), was included as an option for higher performance in the form of range at the 5.5 and 11 Mbps rates, as it provided for a 3 decibel (dB) coding gain.
The second offshoot of 802.11 was designated as 802.11a. It ventured into a different frequency band, the 5 GHz U-NII band, and was specified to achieve data rates up to 54 Mbps. Unlike 802.11b, which is a single carrier system, 802.11a utilized a multi-carrier modulation technique known as orthogonal frequency division multiplexing (OFDM). By utilizing the 5 GHz radio spectrum, 802.11a is not interoperable with either 802.11b, or the initial 802.11 WLAN standard.
In March 2000, the IEEE 802.11 Working Group formed a study group to explore the feasibility of establishing an extension to the 802.11b standard for higher data rates greater than 20 Mbps. In July 2000, this study group became a full task group, Task Group G (TGg), with a mission to define the next standard for higher rates in the 2.4 GHz band. The new standard was which came into existence in July 2003 is named as IEEE 802.11g and operates into 2.4 GHz ISM band and the data rate is up to 54 Mbps.
ECE Department, GITAM University
For basic rate of 1 and 2 Mbps, IEEE 802.11g uses DSSS scheme, for 5.5 and 11 data rates IEEE 802.11g uses CCK encoding scheme, and for 6, 12 and 24 Mbps data rates IEEE 802.11g uses OFDM scheme. Theses all are mandatory modes for IEEE 802.11g PHY. Table 2.1 shows all the mandatory and optional modes for IEEE 802.11b, IEEE 802.11g, and IEEE 802.11a. Table 2.1 Data Rates for All Modes
Mbps multi carrier 1 2 5.5 6 9 Single Single Single Multi Multi Barker Barker CCK PBCC Barker Barker CCK OFDM PBCC CCK-OFDM OFDM, CCK-OFDM 11 12 18 Single Multi Multi CCK PBCC CCK OFDM PBCC CCK-OFDM OFDM, CCK-OFDM 22 24 33 36 Single Multi Single Multi OFDM PBCC CCK-OFDM PBCC OFDM, CCK-OFDM 48 Multi OFDM, CCK-OFDM 54 Multi OFDM,CCKOFDM OFDM OFDM OFDM OFDM OFDM OFDM OFDM OFDM
802.11g achieves the high 54 Mbps data rates (Table 2.1) of 802.11a in the 2.4 GHz band thereby maintaining compatibility with installed 802.11b equipment.
2.2.1.3 Infrared
The Infrared PHY utilizes infrared light to transmit binary data either at 1 Mbps (basic access rate) or 2 Mbps (enhanced access rate) using a specific modulation technique for each. For 1 Mbps, the infrared PHY uses a 16-pulse position modulation (PPM). The concept of PPM is to vary the position of a pulse to represent different binary symbols. Infrared transmission at 2 Mbps utilizes a 4 PPM modulation technique. This specification was designed for indoor use only.
even eliminate ISI because the delay spread will then be shorted than the symbol period. This also eliminates the needs for a complex multi-tap time domain equalizers [17]. OFDM actually combine the data and transmit them in block. The size of each block is determined by the number of sub-carriers used to convert the serial stream of data to parallel stream
demodulation in order to achieve orthogonality. Since DFT has heavy computational requirements, therefore, Fast Fourier Transform (FFT) was utilized. For an N point discrete Fourier Transform the required number of computations is N2, but that for FFT is Nlog (N), which is much lesser than DFT. In this way the problem of bandwidth inefficiency due to the placement of guard bands between sub-channels was solved and a new technique Orthogonal Frequency Division Multiplexing came into being. As OFDM is a multi-carrier modulation technique, therefore, the input data is split and mapped onto different sub-carriers. Each carrier is modulated using one of the singlecarrier modulation techniques discussed above. The OFDM system successfully avoids any inter-channel interference (ICI) because the carriers are kept orthogonal. In addition, a cyclic prefix (CP) is added before the start of each transmitted symbol to act as a guard period preventing inter-symbol interference (ISI), provided that the delay spread in the channel is less than the guard period . This guard period is specified in terms of the fraction of the number of samples that make up a symbol.
11
fading which OFDM is able to overcome by utilizing its parallel, slower bandwidth nature. This makes OFDM ideal to handle the harsh conditions of the mobile wireless environment. The introduction of cyclic prefix made OFDM system resistant to time dispersion . OFDM symbol rate is low since a data stream is divided into several parallel streams before transmission. This make the fading is slow enough for the channel to be considered as constant during one OFDM symbol interval. Cyclic prefix is a crucial feature of OFDM used to combat the inter-symbol interference (ISI) and inter-channel-interference (ICI) introduced by the multi-path channel through which the signal is propagated . The basic idea is to replicate part of the OFDM timedomain waveform from the back to the front to create a guard period. The duration of the guard period should be longer than the worst-case delay spread of the target multipath environment. The use of a cyclic prefix instead of a plain guard interval, simplifies the channel equalization in the demodulator. In wire system, OFDM system can offer an efficient bit loading technique It enables a system to allocate different number of bits to different sub channels based on their individual SNR. Hence, an efficient transmission can be achieved. One of the major disadvantages of OFDM is its requirement for high peak-to average power ratio (PAPR). This put high demand on linearity in amplifiers. Second, the synchronization error can destroy the orthogonality and cause interference. Phase noise error and Doppler shift can cause degradation to OFDM system . A lot of effort is required to design accurate frequency synchronizers for OFDM. OFDMs high spectral efficiency and resistance to Multipath make it an extremely suitable technology to meet the demands of wireless data traffic. This has made it not only ideal for such new technologies like WiMAX and Wi-Fi but also currently one of the prime technologies being considered for use in future fourth generation (4G) networks.
12
standard and for terrestrial digital video broadcasting (DVB) .In fixed-wire applications, OFDM is employed in asynchronous digital subscriber line (ADSL) and high bit-rate digital subscriber line (HDSL) systems. It has been proposed for power line communications systems as well due to its resilience to dispersive channel and narrow band interference. It has been employed in WiMAX a well.
13
14
15
CP Insertion: Copies some samples from the end of the symbol to the front to add some redundancy to the symbols. These duplicated samples are known as a cyclic prefix (CP). The purpose of the cyclic prefix is to avoid Inter-Symbol Interference (ISI) caused by multipath propagation. This block also adds a preamble before the first transmitted symbol. A preamble is a collection of predefined complex numbers known by the receiver so that it can detect the start of new transmission. The preambles for the two protocols have similar structure. After CP insertion, the symbol are converted into analog signals by D/A converter and transmitted through the air.
16
Channel Estimator: Uses the information from pilots to estimate and compensate for frequency-dependent signal degradation. The channel estimator estimates and corrects the errors caused by multipath interference. Similar to the synchronizer, there are many different algorithms for channel estimation. Many of them use either the preambles or the pilots to estimate the effect of the interference on each data subcarrier. We parameterize the channel estimator by protocol-specific preamble and pilot values. Demapper: Demodulates data and converts samples to encoded bits, which are used by the FEC decoder. The number of encoded bits generated per sample is determined by the specific modulation scheme. The parameters of this block are modulation schemes supported and the functions for converting samples to decisions. Deinterleaver: Reverses the interleaving performed by transmitter and restores the original arrangement of bits. FEC Decoder: Uses the redundant information that was introduced at the transmitter to detect and correct any errors that may have occurred during transmission. Both 802.11a and 802.16 use the Viterbi algorithm [13] to decode convolutionally encoded data. To support multiple protocols, the decoder uses the same parameter settings as the convolutional encoder at the transmitter side. Since 802.16 also uses Reed-Solomon encoding, corresponding Reed-Solomon decoder that supports appropriate profiles is used in the receiver side. Descrambler: Reverses the scrambling performed by the transmitter. RX Controller: Based on the decoded data received from Descrambler, the RX Controller generates the control feedback to S/P block.
17
Literature survey of complete Physical layer design of OFDM transmitter and receiver
Creating a top level design of the complete system Determining the basic operation of each block and creating the appropriate logic I/O integration of the various logic blocks Description of design functionality using Verilog hardware description language CADENCE and XILINX are used to simulate the design functionality and to report errors in desired behavior of the design
Synthesis of the defined hardware is done which includes slack optimization, Power optimizations followed by placement and routing
18
4. Scrambler
Overview of Scrambler
2.1 Introduction of Serial Scrambler Scramblers are a class of substitution ciphers and have been found to be suitable for various security requirements such as those used by cable and satellite TV operators and mobile phone service providers. A Scrambler is a coding operation which basically randomizes the data streams. In addition to its use as a stream cipher, a scrambler is commonly used to avoid long strings of 0s and 1s which are responsible for DC wander and synchronization problems in communication circuits. Scramblers are very popular for their use to encrypt video and audio signals for broadcasting and many other applications. The low cost and complexity, high speed of operation and easy to use are the main features of scramblers. The scrambler is built as a shift register with stages depending on the given characteristic equations. There is feedback taken from stages according to the polynomials listed in the characteristic equation. On the transmit side the scrambler may be started with a preload of a specific data content. This is important later on when we talk about testing. This preload is also called seed. With a known seed and a known input pattern the scrambler output is deterministic, which means the output of the scrambler can be calculated. On the receive side the de-scrambler synchronizes automatically from the incoming data. Once pretty obvious data meaning a sequence of data with known a pattern and easily recognizable is sent through a scrambler, it is no longer meaningful to the human eye. Even a repetitive pattern will no longer be periodic as the scrambling adds the total history to the current data stream. So any periodic or specifics inside the patterns cannot be seen inside a scrambled stream. It needs to be de-scrambled again. At present high-speed transmission or networking specifications, there are serial scramblers in it. Serial scrambler consists of several registers and modulo-2 adders (XOR gates), and interconnected in serial form. Its architecture consists of a shift
register with feedback path which perform modulo-2 addition using the outputs of some registers, which is also called Linear Feedback Shift Register (LFSR), Pseudo-Random Bit Sequences (PRBS) generator and Pseudo noise (PN) generator, etc.. The
19
scrambler produces scrambling sequences to perform the operation of modulo-2 addition with input data. The characteristic of scrambler depends on its interconnection. That is to say different feedback connections will produce different output sequences. The output sequences are pseudo-random, which means the sequences are not really random data. It repeats the same sequences after one period, but output sequences within the period are very close to random data. Thus the period is longer, the output sequences are more random. The period of scrambler output sequences is determined by the number of shift register and feedback period of taps. A N-stage scrambler has the maximal
called Maximal Length Sequences (m-sequences).For example, the period of a 3-stage scrambler, as shown in Fig. 2-2(a), has maximal period 7 ( ), but the period of Fig.
2-2(b) is only 4. The number in the bracket is the output of register that connects to the XOR gate. If the feedback taps connect improperly, it would not produce msequences and its period is shorter. We can describe scrambler by a mathematic equation, which is called Characteristic Polynomial. expression For instance, scrambler of IEEE802.3ae has polynomial ( ) (2-1)
Eqn. (2-1) means that the output of the 7th and 6th register performs XOR operation and the result of operation is fed back to the first register as input, as shown in Fig. 2-3. Whenever the transmission frame starts, all registers of serial scrambler reset to initial condition and then scrambles with input data repeatedly. By this operation, the scrambled output data becomes pseudo-random data and transmitted to the channel. The probability of 1 or 0 of the scrambled output sequences is almost equal ( ( ) ( ) ) and the number of 0 differs from the number of 1 by
( )
20
The Advantages of Scrambling The input data is scrambled by performing the XOR operation with the scrambler output sequences. Thus, the scrambled data becomes pseudo-random data no matter what the original input data is. It is good to transmit pseudo-random data in the channel for the communication system. It has three advantages for transmitting pseudorandom data and using the scrambler circuit, which will be discussed in details in the following section.
2.2.1
There are many channels that are used to transmit data simultaneously in communication system. If one of the channels produces large amounts of noise to interfere other channels, it would be troublesome. Actually, if the channel transmitted repetitive sequences of data, from the spectrum point of view, it could have large
21
amounts of energy concentrated at some frequencies. These large amounts of energy could lead to electromagnetic impulse (EMI) to affect other channels. However, after scrambling, it becomes random data in time domain and the probability of 1 or 0 is equa1. On the other hand, the transmitted energy is spread more uniform across the transmission frequency band and minimizes the EMI damage to other channels. Data Transitions The transmission format of signal can be classified into two categories: Non-Return to Zero (NRZ) and Return to Zero (RZ).The former is that when data is 1 in high level, it will not change level to low level until data is 0. The latter is on the contrary. When the data is 1 in high level, it will change level to low level after half interval of one bit time, as shown in Fig. 2-5.Thus, the RZ format has more transitions than NRZ in the same transmitted data, and it is beneficial for receiver to synchronize with transmitter by these data transitions. However, the bandwidth requirement of RZ is twice that of NRZ. Therefore, scrambler can be used to make the same effect on NRZ format
for synchronization purpose. If the transmitted data bytes or control signals are long term 0 or 1 during
transmission, after scrambling, the scrambled data will have more transitions and become more random. It can avoid the generation of transmitting a constant control signal for long periods. Encryption The scrambled data is the result of XOR operation on the input data and scrambler output sequences, the scrambled data is quite different from the input data. By doing so, the data transmitted in channel is random and correlation of this sequence is very close to zero. It is random data until the receiver is descrambled, so it is like to encrypt the original data, as shown in Fig.2-6. coder and a descrambler is a decoder.\ In other words, a scrambler can be viewed as a
22
Scrambler is designed for the polynomial x7 + x4 + 1 With the initial state 1000000 .
23
Fig 4.1 Scrambler circuit diagram Verilog code for implementation of scrambler : module scramblerr(clk,rst,in,sout,lfsr1); input clk,rst,in; output sout; output[6:0]lfsr1; reg [6:0]lfsr; always@(posedge clk or negedge rst ) begin if(!rst) lfsr=7'b1010000; else lfsr={in,lfsr[6:1]}; end assign sout=lfsr[0]^lfsr[4]^in; assign lfsr1=lfsr; endmodule
Descrambler
24
Verilog code for implementation of descrambler: module descrambler(clk,rst,in,desout,lfsr2); input clk,rst,in; output [6:0]lfsr2; output desout; reg [6:0]lfsr3; reg desout1; always@(posedge clk or negedge rst) begin if(!rst) lfsr3=7'b1010000; else begin desout1=in^lfsr3[4]^lfsr3[0]; lfsr3={desout1,lfsr3[6:1]}; end end assign lfsr2=lfsr3;
ECE Department, GITAM University
25
Verilog code for implementation of datapath of scrambler/descrambler : module completescramb(in,clk,rst,desout,sout,lfsr1,lfsr2); input in,clk,rst; output sout; output desout; output [6:0]lfsr1,lfsr2; wire sout,desout; scrambler s1(clk,rst,in,sout,lfsr1); descrambler ds1(clk,rst,sout,desout,lfsr2); endmodule
Verilog code for testbench: `timescale 1ns / 1ps module test1; reg in, clk, rst; wire desout, sout; wire [6:0] lfsr1,lfsr2; completescramb u1 (in,clk,rst,desout,sout,lfsr1,lfsr2 ); initial begin clk=1'b1; rst = 1'b0; in=1'b0; #10 rst = 1'b1; end initial begin forever #5 clk=~clk;
ECE Department, GITAM University
26
end initial begin # 5 in = 1'b1; # 50 in = 1'b1; # 55 in = 1'b1; # 70 in =1'b 0; # 90 in = 1'b0; # 120 in = 1'b0; # 150 in = 1'b0; # 170 in = 1'b0; # 190 in = 1'b0; # 220 in = 1'b1; # 250 in = 1'b0; # 270 in = 1'b1; # 290 in = 1'b0; # 300 in = 1'b1; # 320 in = 1'b1; # 350 in = 1'b1; # 380 in = 1'b1; # 3000000 $finish; end endmodule
27
The Parallel Scrambler: Parallel scramblers are used to achieve high throughput. In IEEE 1394b standard, the input to the transmitter is 8 (or 16) in parallel. So the input works at byte time (8Ts), where Ts is the bit time. But serial scramblers produce output sequences only one bit in a clock cycle. Therefore, it transforms serial scrambler output to parallel outputs by bitshift registers, as shown in Fig. 2-7. The output of serial scrambler is connected to bit-shift registers, and bit-shift registers shift one bit data from the serial scrambler output at one clock cycle. The eight parallel registers generate parallel scrambler data after bitshift registers shifting eight bits data. It totally needs one byte time to produce eight parallel scrambler outputs, which is the same as serial scrambler that takes eight clock cycles of bit time to generate scrambler output bit by bit.
Thus, due to the technology limitation, it cannot work at very high scrambling rate. Thus, a new architecture and algorithm i.e parallel scrambler, which can overcome the disadvantage mentioned above has been developed. In general, multiples of base rate signals are multiplexed and then scrambled before transmission
ECE Department, GITAM University
28
and then is descrambled and demultiplexed after reception in the receiver end, as shown in Fig. 3-1(a). Scrambling used to be done serially or be done as paralleling bit-shift outputs, mentioned in section 2-3. Nevertheless, as the operating frequencies of transmission systems grow beyond Gbps, serial scrambling techniques were no longer applicable in these specifications. For example, in 10 Gbps Ethernet or 40 Gbps fibre transmission, with serial scrambling method, this would mean the scrambler shall work at frequency of 10/40 GHz which is not feasible with todays silicon-based CMOS integrated circuits. The requirement of high working frequency can be resolved by using parallel scrambling techniques [9-13] to enable the scrambling process at the lowfrequency base rate, as shown in Fig. 3-1(b). Under parallel scrambling, sets of scrambling processes are performed at the base rate simultaneously, which collectively achieves the effect of serial scrambling when the scrambled base-rate signals are multiplexed to form a
transmission-rate signal.
is that the number of inputs of the modulo-2 adders (XOR gates) used in the feedback loops of the pseudorandom code generator is more than two for some parallel ports. However, having to process the modulo-2 additions of more than two inputs will lead to an increase in the processing delay and lower the maximum working rate. Moreover, in todays deep sub-micron CMOS process, number of fan-outs and interconnection length also become a significant factor that affect the processing delay. Thus, the architecture shall be regular and have less fan-outs in the critical path.
29
30
Figure 4.4 Convolutional Encoder Figure 5.1 shows the I/O parameters of the Convolutional Encoder. Input bus is 1 bit wide and arst_n is the asynchronous reset input. A negative edge on the arst_n input resets the encoder. A bit is latched in at the positive edge of the clock. For every input bit there is a two bit wide output designated by even and odd.. Convolutional Encoder can be implemented using either a shift register or by using . However, a shift register gives an easy to implement and area efficient solution. For the configuration of m=1, n=2 and k (constraint length) =7. Figure 5.1 shows how the Convolutional encoder is implemented in the proposed design using a shift register. Initially all zeroes are stored in the register. When the first input bit arrives it is shifted into the register from left and the 2 bit output appears on the lines designated as even and odd. The even output is generated by adding the contents of 1st, 0, 3rd, 4th and 6th stages of the shift register, whereas the odd output is generated by adding the 5th, 0, 3rd, 4th and 6th stages of the register. This addition is modulo-2 addition carried out through XOR gates (modulo-2 addition is basically a XOR operation). Just like the Scrambler the memory elements here are D-flip-flops as well.
31
Verilog code for implementation of Convolutional Encoder: module convolencode(in,clk,rst,out); input in,clk,rst; output [1:0]out; reg [2:0]a; always @(posedge clk or negedge rst) begin if(!rst) a=3'b000; else begin a={in,a[2:1]}; end end assign out[0]=a[2]^a[1]^a[0]; assign out[1]=a[2]^a[0]; endmodule
32
Viterbi decoder is a basic and important block in any Code Division Multiple Access (CDMA), and CDMA system uses forward error correction schemes like convolution encoder to prevent interference. The Viterbi algorithm may be viewed as a solution to the problem of maximum a posteriori probability estimation. One of the main blocks of a modem used during forward link demodulation is a Viterbi decoder. A normal Viterbi decoder trace back firstly and then decoding results. The aim of this paper was to design Viterbi decoder for high throughput. It means that this design can continuously trace back and simultaneously process decoding results. In this way, the Viterbi decoder is more
33
efficient. This has been the focus in the present thesis on convolutional decoder suing viterbi algorithm.Several important design issues have also been discussed in the thesis report, such as organization of convolution encoder and Viterbi decoder, branch metric unit computation method, the design of add compare select, parallel unit of trace back and decoder. Fig.1. The structure of Viterbi decoder BMU (Branch Metric Unit) receives the input signals and compute the difference between the input signals and the expected values, and then outputs the results to the Add Compare Select. ACS (Add Compare Select) completes the survivor paths and generates the decision vector. This module has two comparers, and these two comparers are parallel. The one is used for comparing signal bit and the other is used for comparing data bits. The advantage of this design is considerably reducing the circuit timing delay [7].MEMORY is cycle used for saving the survivor paths [8].The core unit of this design is the TB (Trace Back) [9] and DECODER, which include two function modules. The one is TB that read the information of the survivor paths in the MEMORY. The other is DECODER that computes the output results. These two modules work simultaneously but in different MEMORY address. In this way, the Viterbi decoder is more efficient than a normal one because the DECODER do not need to wait TB. In our implementation, the Viterbi decoder consists of two modules: the path metric unit and the Traceback unit. The path metric unit contains a 2k word memory, where each entry is essentially the probability that a sequence of input bits ended in that state. In practice, cumulative error between the hypothesized bit stream and the received bit stream for that state is used as the path metric for that state. The traceback unit records the most likely n state sequence leading to each state, encoded as one bit per state transition, logically organized as an n entry shift register where each entry is 2k bits. The path metric unit updates all 2k entries for every 2 observations received from the demapper module. Once all the new path metrics are computed, the old path metrics can be discarded. As it computes each new path metric, it records the previous state pointer in the traceback unit. After the path metric unit updates the values for all the states, the traceback unit follows the recorded previous state pointers and emits the bit corresponding to the oldest transition in the sequence.
ECE Department, GITAM University
34
Path Metric Unit: The path metric unit, which is shown in Figure 4, contains one or more Add-CompareSelect (ACS) units for calculating the path metrics for each state. The ACS computes two path metrics at a time, as shown in Figure 5. The number of ACS units, which is a parameter to the path metric unit, controls number of path metrics updated per cycle. The overall structure of the path metric unit is similar to that of a single FFT stage with the FFT butterflies replaced by the ACS units. With the generalized pipelining technique presented in ,we can easily parameterize the design of the path metric unit with the number of ACS units. This parameterization represents a tradeoff between area and power: the area increases as we increase the number of ACS units, while the power decreases.
35
Traceback Unit: The traceback unit contains a n2k bit shift register and a decoder which reconstructs one bit at a time by traversing the most likely state transition sequence. Traversing n = 35 transitions in one cycle leads to long cycle times. To reduce the critical path, we pipelined the traceback unit, with a parameterized pipeline depth. In the pipelined implementation of the traceback unit, a single pipeline stage in the decoder traverses s pointers. There are t such stages, such that s t = n and n = 35. Each pipeline register needs to store one traceback memory column of 2k bits and a current state index. We varied the number of pipeline stages (1, 5, 7 and 35) to analyze various design alternatives. Figure 6 shows the area and power measurements for different pipeline depths.
36
The results reflect the minimum frequency required to support the 54 Mbps bitrate for 802.11a. The figure also shows the number of bits written to the traceback memory per cycle and the complexity of the address decode logic for reading out the bits from the traceback memory, which is the input size of the multiplexing logic to read the data from each element of the shift register. The results show that the 5-stage decodes consumes the least area and least power.
1) Fixed constraint length architecture: 2) Parameterized constraint length architecture: In this architecture the the constraint length can be varied
37
First al individual blocks of the transmitter section is to be done and then later the receiver section is to to be implemented. All the individual blocks are to be integrated to complete the transmitter and the receiver pipelines and are integrated to get the complete design of the physical layer design of 802.11a wireless protocol. After the successful simulations of the design of the physical layer of 802.11a wireless protocol 8is carried out then we implement it on the FPGA, which will result in the fulfillment of the hardware implementation of the physical layer design of the 802.11a wireless protocol. Later Ill study and work on the implement the physical layer of the 802.11g wireless protocol which is similar to the 802.11a protocol. Ill the get the hardware implementation of the 802.11g protocol and ill compare the results of the two protocols. Ill undergo a keen study of the differences in the two protocols and then ill furnish the difference between those two protocols. And Ill find out the other techniques that can be incorporated in the two protocols, to make the wireless protocol more efficient and transmit data with higher data rates. The comparative study of the two protocols will be carried and final report will be submitted on the comparative study of the two wireless protocols.
6. REFERENCES
[1] Indian Journal of Science and Technology Vol.2 No. 10 (Oct 2009) ISSN: 09746846,Research article Secure data communication Bhat et al.Indian Society for Education and Environment (iSee) http://www.indjst.org Indian J.Sci.Technol.41.
38
[2]
Design and Simulation of IEEE 802.11g (WLAN) Based PHY,A dissertation submitted by Raj Kumar Vaishya ,INDIAN INSTITUTE OF TECHNOLOGY, DELHI.
[3]
Design and Implementation of a Parallel Processing Viterbi Decoder Using FPGA Lei -ou Wang 1, Zhe-ying Li 1 iInstitute of Micro-electronic Application Tech,B eijng Union University Beijing, China.
[4]
Multi-Clock Pipelined Design of an IEEE 802.11a Physical Layer Transmitter Maryam Mizani and Daler Rakhmatov University of Victoria Department of Electrical and Computer Engineering 2006 IEEE.
[5]
On the design of an FPGA-Based OFDM modulator for IEEE 802.11a.2nd International Conference on Electrical and Electronics Engineering (ICEEE) and XI Conference on Electrical Engineering (CIE 2005) Mexico City, Mexico. September 7-9, 2005.
[6]
Design and Simulation of IEEE 802.11g (WLAN) Based PHY,A dissertation submitted by Raj Kumar Vaishya ,INDIAN INSTITUTE OF TECHNOLOGY, DELHI.
[7]
On the design of an FPGA-Based OFDM modulator for IEEE 802.11a.2nd International Conference on Electrical and Electronics Engineering (ICEEE) and XI Conference on Electrical Engineering (CIE 2005) Mexico City, Mexico. September 7-9, 2005.
[8]
Multi-Clock Pipelined Design of an IEEE 802.11a Physical Layer Transmitter Maryam Mizani and Daler Rakhmatov University of Victoria Department of Electrical and Computer Engineering 2006 IEEE.
[9]
VHDL modeling and simulation of data scrambler and descrambler for secure data communication G.M. Bhat*, M. Mustafa**, Shabir Ahmad** and Javaid Ahmad.
[10]
VHDL modeling and simulation of data scrambler and descrambler for secure data communication G.M. Bhat*, M. Mustafa**, Shabir Ahmad** and Javaid Ahmad
39
[11]
A Design Manual for Implementation of Projects on FPGAs and ASICs Using Verilog ByDr. S. Ramachandran Indian Institute of Technology Madras, India Digital Integrated Circuits A Design Perspective 2/e Jan M. Rabaey, Anantha handrakasan, Borivoje Nikoli Chapters 6 and 11.
[12]
40