CHAPTER 3 HIGH THROUGHPUT ANALYSIS

Transcription

1 56 CHAPTER 3 HIGH THROUGHPUT ANALYSIS 3.1 INTRODUCTION In the previous chapter, the overall data rate maximization of the OFDM-MIMO system for the a Wireless LAN environment using optimized adaptive resource allocation algorithm was derived. This chapter deals with the analysis and the improvement of the system Throughput for the same environment. The system throughput formula is derived for a standard and compared against the traditional approach to estimate the throughput of the system. The system Throughput is also analyzed for various configurations by varying the number of transmitting and receiving antennas, the applied modulation schemes and the number of subcarriers. Typically, the OFDM-MIMO system employs 64 subcarriers, but the number may vary depending on the system utilization. The comparative study proves that the new derived system Throughput formula with increasing the number of antennas and the selection of modulation scheme improves system Throughput and shrinks the Bit Error Rate (BER). For signal detection at the receiver, the proposed MIMO Detection Algorithm is developed to eliminate the CoChannel Interference (CCI) effect and its BER performance is compared with the existing detection technique. Again, the system Throughput is further developed by combining the Throughput Enhancement scheme with adaptive modulation using Channel State Information (CSI).

2 57 High Throughput system needs the synchronisation between transmitter and receiver which is achieved by the proposed synchronisation estimator technique. Its Mean Square Error (MSE) performance will be evaluated and compared with existing standard Schmidl s estimator. 3.2 LITERATURE SURVEY The WLAN environment is a popular mode by which the transmission data rate demanded by each user has increased. The actual rate at which the information is transmitted through the shared medium is called Throughput. Sebastein Simons et al (2003) presented in detail about the high throughput requirement of WLAN and also discussed the standard value of the data rate of WLAN developed by IEEE working groups. They concluded that the physical layer data rate of has already extended from the original 2Mbps to 11Mbps for b and 54Mbps for a/g. The authors however suggested that this extension is not sufficient and further improvement is needed for a WLAN environment, since WLAN is an emerging wireless network at home, enterprise and public access environment. In the enterprise, high speed WLAN represents a flexible alternative or compliment of wired Ethernet. This provides motivation for a continued increase in the available data rate beyond 10Mbps. In public access scenarios, WLAN has the capability to provide high speed Internet access, requiring an optimal trade-off between bit rates and ranges. Further, the home environment represents a number of significant challenges namely the simultaneous distribution of high speed internet, telephony or high videoaudio displays inside the house. Such applications demand the efficient high throughput WLAN environment. Sampath et al in 2002, Van Zelst and Schenk (2004) have made known that the combined application of multi antenna technology and OFDM

3 58 yields a unique physical layer capable of meeting the challenges posed by the wireless environment. Even in terms of channel capacity, it is found that OFDM-MIMO systems are far superior to other such configurations like SISO (Single input Single Output), SIMO (Single Input Multiple Output) and MISO (Multiple Input Single Output). Simon Haykin (2002) has stated in his book that, if the number of transmitting antennas be M t, the number of receiving antennas be N r, Channel bandwidth be B and the signal at the receiver have an average Signal to Noise Ratio be SNR, then the system capacity of different antenna systems C will be as listed in Table 3.1. Table 3.1 Capacity of the various system configuration System Configuration SISO System SIMO System MISO System MIMO System - same signal over each antenna MIMO System - different signal over each antenna System Capacity B.log 2 (1+SNR) B.log 2 (1+N r.snr) B.log 2 (1+M t.snr) B.log 2 (1+M t.n r.snr) M t.b.log 2 (1+(N r /M t ).SNR) The theoretic capacity approximately increases linearly as the number of antennas is increased. Ng et al (2002) has described the Bell Labs Layered Space Time (BLAST) architecture, which utilizes multiple antenna arrays at both transmitter and receiver to provide high capacity wireless communications in a rich scattering environment. It can be proved, using Shannon s Capacity formula, that the capacity of a MIMO system with different signals transmitted by each antenna increases linearly with respect to the number of transmitting antennas. Hence it can be said that this configuration leads to better system utilization and high throughput.

4 59 To achieve high throughput a suitable MIMO detection technique is considered and classified into four types. The first two methods are based on linear estimation theory called Zero Forcing (ZF) and Minimum Mean Square Error (MMSE) methods, the third one is called Vertical BLAST (VBLAST) method, where one of the linear estimators is combined with feedback network. The last one is a non-linear Maximum Likelihood Detection (MLD) method. Li and Sollenberger s (1998) report shows that OFDM applied to MIMO systems with VBLAST signal processing at the receiver is a practical solution for CCI between transmitted and received substreams of single frame TDMA data. Braswell and McEachen (2001) showed an experimental 4 4 MIMO test-bed for wireless communication systems. Recent technical literature has been reviewed to present some of the basic characteristics of OFDM-MIMO systems analyzed by Li and Cimini (2002) and Li et al (2003) that make them attractive for high data rate transmission over wireless channels and the problems with the CCI associated to multi user operation, which can reduce their performance. A Comprehensive description and complexity analysis of ZF, MMSE, VBLAST and MLD detection schemes are given by Van Zelst in his thesis work published in His research work compares the performance of these detection schemes and antenna configurations on perfect Additive White Gaussian Noise (AWGN) channel, but the channel interference effects are not considered. WLAN environment needs a suitable detection technique for this interference cancellation. Another view to improve the system throughput is selected in terms of channel enhancement scheme. Heath et al (2001) maintained that the system throughput may be improved by the usage of transmission control scheme. Milani et al (2002) furnished a solution in terms of antenna selection based on the received signal SNR. Taira et al (2004) detailed about the

5 60 throughput enhancement scheme in terms of the transmit antenna selection and the adaptive modulation for single carrier system. Suto et al (2004) analyzed the same in optimization technique, subject to adaptive power control using the BLAST system. This is a very good solution, to satisfy the WLAN users need, where high throughput can be achieved by the proper selection of MIMO configuration and its corresponding detection method combined with channel enhancement scheme. Zanella et al (2005) discussed about MMSE reception with Interference cancellation for MIMO system. Chung-Wen Yu and His-Pin Ma (2006) introduced a low complexity scalable MIMO detector. Many research works are carried out in the field of MIMO detector (Jan Wang and Shaoqian Li 2007, Kim et al 2008). In the latest reference, Jinyoung An and Sangchoo Kim (2009) provided a successful interference cancellation scheme in Ultra Wide Band (UWB) MIMO systems. If the system throughput is increased further and further, the subcarriers are spaced closer and closer together in frequency, whereby the spectrum may overlap each other and lead to ICI (Inter Carrier Interference). Therefore an accurate Synchronization technique is thus essential for reliable reception of the transmitted data. Synchronization of the carrier frequency at the receiver must be performed very accurately, or there will be loss of orthogonality between the symbols. OFDM systems are very sensitive to carrier frequency offsets since they can only tolerate offsets which are fractions of the spacing between the subcarriers without a large degradation in system performance. Moose in 1994 introduced the effects of the frequency offset on the OFDM system. He also described a technique to estimate the frequency offset using two repeated OFDM symbols. The Maximum Likelihood Estimation (MLE) algorithm and variance of estimation are also derived, but the estimation range is limited inside half subcarrier interval. Schmidl and Cox

6 61 (1997) recommended a method for time and frequency synchronization respectively. Two training symbols are placed at the beginning of the frame and the first symbol has identical halves in time domain, so that the correlation between these two halves can be carried out to find out the timing metric in the receiver. Yun Hee Kim et al (1999) made an improvement to the above method by using a differentially coded training symbol to find out the integral part of the frequency offset and use only one OFDM training symbol. Morelli and Mengali (1999) proposed the best linear Unbiased Estimation method, where only one training symbol is used, but it consists of many repeated parts. It improved the frequency offset estimation range and accuracy but increased the computation. Minn et al (2000) modified Schmidl and Cox s method and announced two new methods. The first method uses all samples over one symbol period (excluding guard interval) and used it in calculating the half symbol energy required in the timing metric. The timing metrics is averaged over a period or window of guard interval length. The second method uses a training symbol containing four equal length parts but the last two parts have a different sign. These two methods provide smaller estimator variance, but have large MSE in ISI channel. Byungjoon Park et al (2003) assert a novel timing estimation method using a training symbol consisting of four parts. The first two are symmetric and the last two are conjugate of first two respectively. This method produces a significantly smaller MSE than the Minn et al (2000) method. Seung Duk Choi et al (2006) presented a timing offset estimation method and designed a new time domain preamble to give smaller MSE than other previous estimators even in the fast varying channel. Fan Wu and Mose Ali Abu-Rgheff (2007) introduced a tutorial research of estimation synchronization method in terms of frequency offset technique. Hung Nquyen-Le et al (2009) proposed the joint channel Estimation and Synchronization technique in the presence of carrier and Frequency offset.

7 PROPOSED TECHNIQUE FOR ANALYSIS The throughput analysis of OFDM-MIMO System IEEE a specification WLAN environment functions are split into a number of submodules. Initially, at the transmitter side, multi user data streams are demultiplexed into substreams defined as serial to parallel conversion and each substream is mapped by QPSK/QAM or by any proper modulation techniques on the subcarriers. The mapped symbols frequency domain are converted to time domain by IFFT evaluation process and then fed to its corresponding transmit antennas. All these techniques are easily implemented by OFDM process. The subcarrier and bit allocations are done by optimized adaptive resource allocation algorithm to get the maximum data rate on each channel. With the help of MIMO configuration, OFDM symbols are transmitted by M t transmitter antennas and received by N r receiver antennas in the rich scattering environment as shown in Figure 3.1. Input Serial Data OFDM transmitter 1 M t Rich Scattering WLAN Environment 1 N r MIMO Detection, Throughput Enhancement Process & OFDM demapping Output Serial Data Channel State Information Figure 3.1 Configuration of OFDM-MIMO system with CSI At the receiver, the proposed MIMO detection algorithm detects the incoming OFDM symbol and executes the throughput enhancement process. Channel State Information (CSI) is fedback to transmitter to improve the

8 63 throughput of the system. The CSI is used for reducing the amount of error occurring in a channel. With the help of CSI technique the state of the error in a channel can be identified and can be reduced so that SNR can be improved more effectively. Channel tracking can be performed by Maximum Likelihood (ML) method and it has been used to acquire accurate CSI. The investigations on high throughput of OFDM-MIMO system based on the proposed techniques analysis are divided into Function submodules and listed in Figure 3.2. START Proposed Estimate equation for System Throughput Study of throughput and BER Performances Comparative Study of Proposed MIMO Detection Algorithm Proposed System Throughput Enhancement Scheme Proposed Synchronization Estimator Method END Figure 3.2 Functional submodules of proposed system throughput analysis

9 64 In the first module of analysis, a comparison between the traditional and the proposed estimate scheme for derived system throughput was done with different modulations of QPSK, 16QAM and 64QAM. The system throughput may vary by increasing the number of subcarriers and MIMO configuration is verified in the next module of analysis. The simulation is performed for 4 different configuration of transmitter and receiver antennas, namely 2 2, 2 4, 4 8 and 4 12 with the number of subcarriers of 32, 64, 128 and 256. The modulation methods of QPSK, 16QAM and 64QAM are used. In the third module of analysis, the proposed MIMO detection algorithm BER performance is compared with the existing methods called ZF, MMSE, MLD and VBLAST. In the fourth module of analysis, the system performance is studied in terms of throughput enhancement scheme with adaptive modulation to achieve a very high system throughput. Based on the above techniques, the throughput is increased to a very high value to be capable for WLAN applications. This high throughput WLAN transmission needs the synchronization between transmitter and receiver stations. Therefore, in the last module of analysis, time and frequency synchronization technique on the system are tested by MSE performance and is compared with existing Schmidl s method Proposed System Throughput Estimation Equation In the previous chapter the combination of OFDM with MIMO transmission system improves the channel data rate by properly allocating resource (bit, subcarrier and power) amongst different users in WLAN environment. The derived optimized Adaptive Resource Allocation algorithm maximizes the minimum capacity user data rate and makes high overall data rate R of the channel in terms of number of the bits per subcarrier, while it simultaneously satisfies the requirements of each user s data rate, BER and the total transmit power constraints.

10 The derived equation for optimized overall data rate R of the channel in Chapter 2 is 65 K N R max r (3.1) k 1 n 1 n,k n,k where r n,k denotes number of bits allocated to k th user on n th subcarrier and n,k is the allocation indicator. In this chapter, WLAN overall data rate or throughput can be improved by deriving a new throughput formula for resource allocated multiuser OFDM-MIMO system. The proposed alternative can be compared with the traditional approach in its simulation. Its performance can be studied for varying MIMO configuration, the applied modulation scheme and number of the subcarriers to a variety of input traffic like image and audio data. The system throughput depends on OFDM symbol size, Transmission rate, Received signal power, Received noise power spectral density, Modulation technique adopted at the transmitter, Channel conditions, SNR and BER. According to Schenk et al (2004) system throughput can be computed using the following elementary formula TR R (1 BER) (3.2) where BER denotes the bit error rate and R is the derived overall data rate of resource allocated channel. In this simulation, this parameter is extended to include the Bit Error Rate of the OFDM system. For the analysis select M-ary QAM modulation and its probability of correct detection at the receiver be P c. P c can be expressed in terms of Pe as c e 2 P 1 P (3.3)

11 66 The probability of symbol error of M-ary QAM P e is defined by ' 1 Pe 1 erfc M E N b 0 (3.4) where E b is the signal energy per bit, N 0 is the noise spectral density and M defines the value of M-ary QAM. The probability of symbol error P e is given by 2 Pe 1 Pc 1 1 Pe 2P e (3.5) where it is assumed that ( P e ) 2 is small enough compared to unity. Hence from the above equation becomes 1 Pe 2 1 erfc M E N b 0 (3.6) The transmitted energy in M-ary QAM is variable in that its instantaneous value depends on the particular symbol transmitted. It is therefore more logical to express P e in terms of the average value E av of the transmitted energy rather than bit energy E b, the relation between the two terms may be expressed as follows: E av 2 M 1 E 3 b (3.7) Accordingly, the probability of symbol error is given by, 1 Pe 2 1 erfc M 3Eav 2 M 1 N 0 (3.8)

12 67 The symbol error probabilities for QPSK and 4-QAM are found to be the same. Hence the same result holds good for QPSK, 16-QAM and 64-QAM. The above result for probability of error holds good in the case of a Single Input Single Output System. In order to extend the result to the Multiple Input Multiple Output case, using Shannon s capacity formula as reference and performing a similar mapping to the equation P e, the probability of symbol error in OFDM-MIMO system is given by 1 3 Nr E Pe Mt 2 1 erfc M 2 M 1 M N av t 0 (3.9) In the case of an OFDM based system, the bits are mapped into symbols of length L before transmission. Hence the error probability of an entire OFDM symbol is to be considered. The formula for throughput can be modified as L b TR R (1 BER) (3.10) Based on this improved estimate, the overall system throughput of the OFDM-MIMO system can be expressed by the following formula: 1 3 Nr E TR R 1 Mt 2 1 erfc M 2 M 1 M N av t 0 L b (3.11) This modified formula helps to evaluate the system performance in terms of the number of symbols correctly received in the presence of channel noise. The above result, relates the system throughput to SNR directly. Hence,

13 it is definitely a better approach to determine the throughput of the system, which can be analyzed by the comparative study with traditional approach Comparative Study of System Throughput The comparative plots between the existing and proposed approaches to determine the throughput of a MIMO OFDM system are presented with the following input conditions: Number of subcarriers N = 64 Number of transmit antennas M t = 2 Number of receive antennasn r = 2 Modulation schemes applied: QPSK, 16-QAM and 64-QAM. Nature of input traffic: Image file Throughput is defined in terms of Mbps. Using 64QAM modulation scheme the proposed estimate for system throughput is increased to 72Mbps compared to traditional approach, which has the maximum throughput of 42Mbps, as shown in Figure 3.3. Similarly the system throughput is increased to 54Mbps from 30Mbps and 39Mbps from 10Mbps using 16QAM and QPSK modulation schemes respectively as shown in Figures 3.4 and 3.5. Compared to the traditional approach, the proposed estimate provides a high throughput and 64QAM Modulation produces very high throughput compared to 16 QAM and QPSK. At low SNR value of 4dB, throughput reaches the maximum value and retains it in the constant value for 64QAM, SNR of 6dB for 16QAM, and SNR of 8dB for QPSK respectively. But in the traditional approach, the throughput gets the maximum value and retains in constant at 8dB in 64QAM, 10dB in16qam and 12dB in QPSK. Ideally a high throughput at low SNR is preferable and it is achieved by the

14 69 proposed system Throughput estimation using 64QAM. Based on the results from Figures 3.3 to 3.5, increasing the degree of modulation it is possible to achieve a better throughput. Figure 3.3 Comparative plot for 64QAM modulation scheme Figure 3.4 Comparative plot for 16QAM modulation scheme

15 70 Figure 3.5 Comparative plot for QPSK modulation scheme Comparative Study of System Throughput and BER Performance The system performance of WLAN OFDM-MIMO system is analyzed in two stages. In the first stage, the system throughput is computed by varying the number of subcarriers, applied modulation schemes and the number of transmitting and receiving antennas. In the second stage, system bit error rate is computed by varying same parameters of the number of subcarriers, applied modulation schemes and the number of transmitting and receiving antennas. The computed system throughput values are plotted against the E b /N 0 in db as shown in Figures 3.6 to The BER is computed by finding the mean square value of detected data minus the original input data. The computed values are plotted against the same parameter of E b /N 0 in db as shown in Figures 3.22 to For the investigation of system performance, the number of subcarriers of 32, 64, 128 and 256 with the Number of transmit antennas and the Number of receive antennas for MIMO Configurations of 2 2, 2 4, 4 8 and 4 12 are selected Throughput performance Number of subcarriers: 32

16 71 For a 2 2 configuration, the maximum throughput of 25Mbps, 35Mbps and 70Mbps respectively are obtained and retained at constant value after the SNR of 8dB for QPSK, 16QAM and 64QAM. But for 2 4, 4 8 and 4 12 configurations it achieves the same after 6dB. At low SNR of 2dB, high degree 64QAM modulation performance is appreciable by increasing the number of transmitter and receiver antennas. The throughputs of 20Mbps, 50Mbps, 60Mbps and 68Mbps are achieved at low SNR of 2dB in the increasing configurations of 2 2, 2 4, 4 8 and Figures 3.6 to 3.9 show the throughput performance with respect to E b /N 0 for different MIMO configuration, different degree of Modulation and for a subcarrier of 32. Figure 3.6 Throughput performance M t = 2, N r = 2, N = 32

17 72 Figure 3.7 Throughput performance M t = 2, N r = 4, N = 32 Figure 3.8 Throughput performance M t = 4, N r = 8, N = 32

18 73 Figure 3.9 Throughput performance M t = 4, N r = 12, N = Number of subcarriers: 64 In 2 2 configuration, the maximum throughput of 30Mbps, 45Mbps and 80Mbps respectively are obtained and retained a constant same value after the SNR of 6dB for 64QAM and 8dB for the other two modulations. But for 2 4, 4 8 and 4 12 configurations the maximum throughput is attained and retained after 4dB of SNR. At low SNR of 2dB, high degree of 64QAM modulation performance is substantial by increasing the number of transmitter and receiver antennas. The throughputs of 15Mbps, 45Mbps, 60Mbps and 70Mbps are achieved at low SNR of 2dB in the increasing configurations of 2 2, 2 4, 4 8 and 4 12 which are shown in Figures 3.10, 3.11, 3.12 and 3.13 respectively.

19 74 Figure 3.10 Throughput performance M t = 2, N r = 2, N = 64 Figure 3.11 Throughput performance M t = 2, N r = 4, N = 64

20 75 Figure 3.12 Throughput performance M t = 4, N r = 8, N = 64 Figure 3.13 Throughput performance M t = 4, N r = 12, N = 64

21 Number of subcarriers: 128 Figures 3.14 to 3.17 show the throughput performance, when the subcarrier takes a value of 128. In 2 2 configuration, QPSK, 16QAM and 64QAM get the maximum throughput of 40Mbps, 55Mbps and 90Mbps respectively and retained at a constant value after the SNR of 8dB. However for 2 4, 4 8 and 4 12 configurations the maximum throughput is achieved and retained after 10dB. At low SNR of 2dB, high degree 64QAM modulation presentation is considerable by growing the number of transmitter and receiver antennas. The throughputs of 20Mbps, 35Mbps, 55Mbps and 75Mbps are achieved at low SNR of 2dB in the increasing configurations. Figure 3.14 Throughput performance M t = 2, N r = 2, N = 128

22 77 Figure 3.15 Throughput performance M t = 2, N r = 4, N = 128 Figure 3.16 Throughput performance M t = 4, N r = 8, N = 128

23 78 Figure 3.17 Throughput performance M t = 4, N r = 12, N = Number of subcarriers: 256 Using the number of subcarrier of 256, in 2 2 configuration, QPSK, 16QAM and 64QAM get the maximum throughput of 55Mbps, 65Mbps and 100Mbps respectively and retained at constant value after the SNR of 8dB. Yet for 2 4, 4 8 and 4 12 configurations the maximum throughput is achieved and retained after 6dB of SNR. At low SNR of 2dB, high degree 64QAM modulation recital is noticeable by raising the number of transmitter and receiver antennas. The throughputs of 30Mbps, 40Mbps, 50Mbps and 80Mbps are achieved at low SNR of 2dM in the increasing configurations. These results are presented in Figures 3.18 to 3.21.

24 79 Figure 3.18 Throughput performance M t = 2, N r = 2, N = 256 Figure 3.19 Throughput performance M t = 2, N r = 4, N = 256

25 80 Figure 3.20 Throughput performance M t = 4, N r = 8, N = 256 Figure 3.21 Throughput performance M t = 4, N r = 12, N = BER Performance

26 Number of subcarriers: 32 Using the number of subcarrier of 32, in 2 2 configuration, QPSK, 16QAM and 64QAM get between 10 0 and If the MIMO configuration is increased, BER performance becomes appreciable and gets the minimum value. In the 4 12 configuration 64QAM modulation produces BER below At low SNR, a high degree of 64QAM performance is substantial by escalating the number of transmitter and receiver antennas. These results are furnished in Figures 3.22 to Figure 3.22 BER performance M t = 2, N r = 2, N = 32

27 82 Figure 3.23 BER performance M t = 2, N r = 4, N = 32 Figure 3.24 BER performance M t = 4, N r = 8, N = 32

28 83 Figure 3.25 BER performance M t = 4, N r = 12, N = Number of subcarriers: 64 Using the subcarrier of 64, in 2 2 configuration, QPSK, 16QAM and 64QAM get between 10 0 and If the MIMO configuration is increased, BER act becomes sound and gets the minimum value. In 4 12 configuration 64QAM modulation, the BER falls below Approximately its response is similar to the number of subcarrier 32. At low SNR, high degree of 64QAM, the performance is noticeable by the growing number of transmitter and receiver antennas. These results are given in Figures 3.26 to 3.29.

29 84 Figure 3.26 BER performance M t = 2, N r = 2, N = 64 Figure 3.27 BER performance M t = 2, N r = 4, N = 64

30 85 Figure 3.28 BER performance M t = 4, N r = 8, N = 64 Figure 3.29 BER performance M t = 4, N r = 12, N = 64

31 Number of subcarriers: 128 Using the number of subcarriers of 128, in 2 2 configuration, QPSK, 16QAM and 64QAM get between 10 0 and If MIMO configuration is increased, BER recital becomes fine and gets the minimum value. In 4 12 configuration, the 64QAM modulation makes the BER below At low SNR, high degree 64QAM performance is made substantial by mounting the number of transmitter and receiver antennas. These results are presented in Figures 3.30, 3.31, 3.32 and 3.33 for the configurations of 2 2, 2 4, 4 8 and 4 12 respectively. Figure 3.30 BER performance M t =2, N r = 2, N = 128

32 87 Figure 3.31 BER performance M t =2, N r = 4, N = 128 Figure 3.32 BER performance M t =4, N r = 8, N = 128

33 88 Figure 3.33 BER performance M t =4, N r = 12, N = Number of subcarriers: 256 Using the number of subcarrier of 256, in 2 2 configuration, QPSK, 16QAM and 64QAM get between 10-1 and If the MIMO configuration is increased, BER performance becomes good and gets the minimum value. In 4 12 configuration 64QAM modulation creates BER below At low SNR, high degree 64QAM performance is made significant by escalating the number of transmitter and receiver antennas. Compared to the previous subcarrier selection, 256 subcarriers with 4 12 configuration turns out to be a satisfactory performance of the system. These results are shown in Figures 3.34 to 3.37.

34 89 Figure 3.34 BER performance M t =2, N r = 2, N = 256 Figure 3.35 BER performance M t =2, N r = 4, N = 256

35 90 Figure 3.36 BER performance M t =4, N r = 8, N = 256 Figure 3.37 BER performance M t =4, N r = 12, N = 256

36 91 OFDM-MIMO system employing 64QAM modulation scheme gives the best response both in terms of Throughput and BER. Increasing the number of antennas proportionally maximizes the throughput and minimizes BER. Also, system response is enhanced when more subcarriers are employed. Increase in subcarriers produce good BER performance. If the subcarrier is less, it should carry more number of bits and it generates a possibility of high BER. But if the number of subcarriers is increased for the same input, each subcarrier will carry only less number of bits, so the possibility of getting a reduced BER. For 2 2 and 4 12, the subcarrier is the same, only the M t and N r values are increased. This produces a minimum BER and High Throughput. Investigation of the potential of MIMO system for providing an increased throughput for WLAN environment is also achieved by proper selection of the detection method discussed in the next section below Proposed MIMO Detection Algorithm By the resource allocation technique, the users data are sent in parallel over multiple antennas, the effective transmission rate is increased in proportion to the number of transmitter antennas used. For simplicity, flat fading is assumed, and based on the number of transmitting and receiving antennas, the matrix channel transfer function is defined as H, where h i,j is the (complex) transfer function from transmitter j to receiver i, and M t N r. Each receiver receives the signals transmitted from all the M t transmitter antennas. When decoding one transmitter antenna, other antennas are treated as interference. The interference problem called CCI can be cancelled by the proposed MIMO detection algorithm. Its BER performance can be compared with the existing popular detection algorithms like ZF, MMSE, MLD and VBLAST techniques.

37 Let s = (s 1, s 2... s Mt ) T denote the vector of transmit symbols, then the corresponding received vector is 92 r Hs w (3.12) where w is a noise vector with components drawn from wide-sense stationary processes with variance 2, and H is the channel Matrix. One way to perform detection for OFDM-MIMO system is by using conventional Adaptive Antenna Array techniques by linear combinatorial nulling method. Conceptually, each substream in turn is considered to be the desired signal and the remainders are considered as "interferers". Nulling is performed by linearly weighting the received signals so as to satisfy some performance-related criterion. For example, consider the conventional two antenna OFDM-MIMO system. In the receiver, because of two antennas receiving the signals composed of user1 and user 2 data, the signal combining between multiple antennas and the interference suppression are performed to divide into two data sequences. The reception signal at the receiver antenna i for n th subcarrier is r h h s w n n n n r h h s n n n n w 2 (3.13) where n is the subcarrier number, h n ij is the branch gain from the transmitter antenna j to the receiver antenna i, s n i is the transmitted signal from the antenna i, w j is the noise at the receiver antenna i and branch means the channel from one transmitter antenna to one receiver antenna. By the proposed MIMO detection algorithm, to estimate the transmitted signal s n i, the inverse matrix V of the channel matrix H = [h ij ] be denoted by

38 93 V v v v n n v n n (3.14) The inverse matrix V applied to the reception signal vector to obtain an estimate of s ij be v v r s v v w n n n n n n v v r s v v n n n n n n w 2 (3.15) The nulling can be performed by choosing weight vectors v i, i = 1, 2,..., M t, such that v i T (H) j = ij (3.16) where M t is the number of transmitting antenna, (H) j is the j-th column of H, and is the Kronecker delta. Thus, the decision statistic for i th substream is d i = v i T. r i (3.17) This linear nulling approach is viable, but superior performance is obtained if nonlinear techniques are used. One particularly attractive nonlinear alternative is to use symbol cancellation as well as linear nulling to perform detection. Using nonlinear symbol cancellation, interference from already detected components of s is subtracted from the received signal vector, resulting in a modified received vector in which effectively fewer interferers are present. When symbol cancellation is used, the order in which the components of s are detected becomes important to the overall performance of the system. The detection processes with CCI cancellation consist of two basic processes such as Interference cancellation process and optimal detection process.

39 Interference cancellation process Initially, CCI cancellation and detection function are carried on by the simple function of nulling technique. The received vector can be written as r = s 1 h 1 + s 2 h 2 + s 3 h s Mt h Mt + w (3.18) where s i is the transmitted symbol from the i th transmitter antenna and h i is the i th column of H. The nulling vector v i is chosen such that v T i h i 0 fo r i j 1 fo r i j (3.19) where () T indicates transpose. Then, using equation (3.19), the decision statistic for the i th symbol is symbol d i v r T i s v h s v h s v h... s v h v w T T T T T 1 i 1 2 i 2 3 i 3 Mt i Mt i x... 0 w i i (3.20) A decision now can be made on d i to estimate the transmitted s Q d (3.21) i i where Q() is the decision function. The effect of symbols already detected can be subtracted from the symbols yet to be detected. This improves the overall performance when the order of detection is chosen carefully. Denoting the received vector r by r 1, if the nulling vector is v 1, then the decision statistic for the first symbol d 1 can be written as T d1 v1 r 1 (3.22)

40 If s 1 = Q(d1) is the estimated s 1 after the decision, then the interference due to s 1 on the other symbols can be subtracted by 95 r2 r1 s1h 1 (3.23) The next symbol is then detected by finding v 2 and then making a decision on v 2 T r 2 and so on. The performance of this successive cancellation and detection scheme depends on the decision taken at each stage Optimal detection order To minimize error propagation the strongest symbols are detected first. Then simultaneously the signals from all the receiving antennas are received. The strongest substream from the received signal is extracted first and then the remaining weaker signals are recovered. This is known as the optimal detection order. A simple optimal ordering is based on the post detection SNR of each substream.

41 96 The SNR for the i th detected symbol of vector r is given by SNR E s 2 i i 2 2 vi (3.24) where 2 is the noise power and E{} denotes the expectation. As v T i h i 2 = v i 2 h i 2, it is seen that a smaller v i 2 value requires the corresponding h i having higher 2-norm. So the SNR for the i th substream is proportional to the norm of the i th column of H. Thus, the optimal detection order is in decreasing order of the 2-norm of the columns of H. The information feed back is used to transmit signals from each antenna. There are huge performance gains in terms of symbol error probabilities when antenna diversity is used. Diversity is one of the most effective ways to combat deep fades. Assuming that the receiver is provided with multiple replicas of the same information bearing signal, and denoting the probability that the instantaneous SNR is below the receiver threshold on a single diversity branch by t, if the receiver is provided with L replicas that fade independently, then the probability that all the branches are at, or below the threshold at the same time is equal to tl. The nulling vector is computed by the following procedure. pseudoinverse of H The nulling vector v i is unique and is the i th row of the v H (3.25) T i i where i denotes the i-th row and + denotes the pseudoinverse. With the Interference cancellation process and decoding, v T i is chosen as the i th row of pseudoinverse of H whose 1 to i 1 columns are set to zero. With optimal

42 ordering, if {q 1, q 2 q Mt } denotes the optimal order, at the q i th stage the nulling vector v T qi is 97 v H (3.26) T qi qi 1 qi where H qi 1 denotes the matrix obtained from H by zeroing the columns q 1, q 2 q Mt Detection algorithm steps Stage 1: Initialization In this stage, the required counters are set and the variables are initialized as depicted 1. Initialize the loop counter i as 1 2. Assign the value of the pseudoinverse of channel matrix H to G 1. G 1 = H + (3.27) 3. Determine the optimal order q 1 as the minimum of the square of G 1. q 1 = min j ( G 1 2 ) (3.28) Step 2: Iteration required results. This is the iterative stage which has to be repeated to obtain the 1. Compute nulling factor v qi T =<G i > qi (3.29) 2. Determine decision statistics d ki as the product of v qi T and r i 3. Estimate transmit vector s ki as s qi = Q(d qi )

43 98 4. Determine next received vector r i+1 = r i s qi H qi 5. + Compute G i+1 = H i+1 6. Determine q i+1 as in the initialization phase as the min j ( G i+1 2 ) 7. Proceed to the first step again, after incrementing the loop counter, i by 1 and stop the iteration, when i = M t. At the receiver, the signal at each receiver antenna will have signal from all the transmit signal coming from different channels. After FFT and the cyclic prefix elimination, the channel frequency response will be the channel matrix of H. The receive vector r will be the channel matrix multiplied by the transmit vector s. Then the proposed detection algorithm will execute the detection, performed on the received vector and the separate transmit signal. Multipath remains an advantage for a OFDM-MIMO system since the frequency selectivity caused by the multipath involves the order distribution of the channel matrix H, thereby increasing capacity. This development is needed to improve the overall throughput of the system. In the simulation, the proposed detection algorithm BER performance is compared with the existing detection algorithms Comparative study In this comparative study, Figure 3.38 shows the BER performance as a function of SNR for WLAN OFDM-MIMO system with 256 subcarrier, 64QAM modulation and 2 2 configuration. This exhibits the data rate of 70Mbps. High data rate communication surely produces channel interference.

44 99 Figure 3.38 Comparative plots for detection techniques with 64QAM for M t =2 and N r =2 At the receiver, different MIMO detection algorithms namely, ZF, MMSE, MLD, VBLAST and the proposed MIMO Detection Algorithms are selected for the performance evaluation, where the proposed detection algorithm has the CCI cancellation concept with detection technique and the results are compared. The same procedure is repeated for 2 4 and 4 12 configurations and exhibits the throughput of 100Mbps by the overall system throughput formula, the BER performances as shown in Figures 3.39 and It can be concluded from Figures 3.38 that for the regarded SNR range, the proposed detection algorithm performs well. ZF detection algorithm requires an SNR which is higher than MMSE detector to reach the same BER performance. MLD and VBLAST perform even better than ZF and MMSE detections. For BER of 10-4, the corresponding SNR for ZF, MMSE, MLD,VBLAST and proposed techniques are 7dB, 8dB, 9dB, 12dB and 17dB respectively. In Figure 3.39 the plots ZF, VBLAST and MLD performance are similar and MMSE detection attains the same at high SNR. The proposed method produces fine response apart from other algorithm in terms of BER

45 100 performance. From Figure 3.40 it can be inferred that in a 4 12 configuration at low SNR of below 20dB the proposed algorithm produces improved BER response. Figure 3.39 Comparative plots for detection techniques with 64QAM for M t =2 and N r =4 Figure 3.40 Comparative plots for detection techniques with 64QAM for M t =4 and N r =12

46 101 By adding extra transmitter and receiver antennas, it can be concluded that the detection algorithm BER performance with respect to E b /N 0 is better and hence preferable to others. In practice, WLAN throughput may extend to very high value. The proposed MIMO detection algorithm throughput performance capability is tested in terms of BER versus E b /N 0 as shown in Figure The proposed detection algorithm is experienced for 70Mbps, 100Mbps and 162Mbps throughput values. Figure 3.41 BER performance of proposed MIMO detection algorithm using different throughputs From the observation, the proposed technique produces a fine response and it has proved that detection algorithm is suitable for high data rate transmission. If the throughput values are increased better BER performance are produced in proportional higher value of SNR. BER of 10-4 is achieved at 10dB of SNR by 70Mbps signal throughput and similarly for other throughput signals 14dB and 19dB of SNR are reached. From the observation, the proposed MIMO detector technique improves the throughput to 162Mbps. The system throughput can be further improved efficiently by throughput enhancement method. This can be implemented by the technique

47 102 of the adaptive modulation methods based on the system SNR. The derived system Throughput, the proposed detection algorithm and the throughput enhancement methods provide the best solution for the system throughput of WLAN environment Proposed Throughput Enhancement Scheme In recent times a higher data rate or higher throughput has attracted a lot of attention, but it generates the complexity in the evaluation of multi channels. This complexity problem can be solved by the new technique enhancement scheme. The complexity reduction scheme and the evaluation complexity are widely investigated by researchers who provide the suggestion that throughput enhancement is possible by the antenna selection and adaptive modulation techniques. In this work, a part of the throughput enhancement is automatically updated by the implementation of Resource Allocation technique described in chapter 2. Based on the bit allocation on each subcarrier, the corresponding modulations are carried out by the Adaptive Modulator at the transmitter. This concept leads to the maximization of data rate, which results in throughput enhancement. Future improvement on throughput enhancement scheme is also proposed by combining the same adaptation Modulation with channel estimation based on evaluated SNR at the Receiver. Channel Estimation: The modulated orthogonal sequences are transmitted in each channel. The channel gains of each subcarrier between transmitter and receiver antennas are estimated at the receiver as defined in proposed MIMO detection algorithm procedure. Calculation of Average SNR: The average SNR in each subcarrier is SNR av 1 N N j 1 SNR j (3.30)

48 103 where, N denotes the number of subcarriers. Generally, under the fading environment, burst error occurs when the received SNR reduces. Errors in the received signal increases when the received SNR is less than the average SNR. If the received SNR is less, then no signal is transmitted from the transmitter antenna. Adaptive Modulation Selection and throughput calculation: The proper modulation method is selected to each channel based on the average SNR and the instantaneous system throughput is calculated. This information is fed back to transmitter to adopt the modulation. This transmission control scheme is implemented by CSI technique. The flow chart of the proposed scheme is shown in Figure START Channel Estimation for all branches Calculate SNR for each substream No To All? Yes Evaluate average SNR and Adopt Modulation scheme No To All? Yes Compute system Throughput and BER performance END Figure 3.42 Flow chart of the throughput enhancement technique

49 104 In the proposed method, the received SNR at each channel can be estimated by the noise power and the branch gain between transmitter antennas and the receive antennas are observed at the receiver. Based on the instantaneous throughput maximization criteria with the help of CSI, the next higher degree modulation is recommended at the transmitter. The scope for further research in high throughput improvement is possible with Cyclic Prefix (CP) Adaptation Technique using Frequency Domain Channel Length Indicator (FCLI) algorithm. The derived System Throughput estimate formula, the proposed detection algorithm and the Throughput enhancement methods provide the best solution for the System Throughput for the WLAN environment. Based on Resource allocation Throughput enhancement, the increase in data rate on each channel or overall throughput of the OFDM-MIMO system fix more number of samples per OFDM frame and it proportionally increases the number of subcarriers. The subcarriers are spaced closer and closer together in frequency domain when the number of subcarriers is increased to provide a better data rate considering that the available bandwidth is the same. This brings the need for stricter synchronization in the system Comparative study The conventional scheme always uses the fixed modulations of QPSK, 16QAM and 64QAM to all transmitter antennas. The system throughput estimated can be improved effectively more by the proposed throughput enhancement otherwise called transmission control scheme achieved by Adaptive modulation technique depending on the SNR and the number of bits on the subcarrier. In Chapter 2 Resource Allocation technique identifies minimum capacity user and maximizes the overall data of the system by increasing the bits on user s subcarrier using the suboptimal resource allocation algorithms and modulates the subcarriers with adaptive

50 105 modulation depending on the number of bits on each subcarrier. Adaptive modulation can select any degree of modulation depending on the given constrains at the transmitter. In the proposed Throughput Enhancement scheme, the WLAN throughput is further improved by CSI transmission control to transmitter about the selection of the Adaptive modulation depending on each evaluated SNR and the instantaneous throughput value. As a result of the combination of the proposed detection algorithm with Adaptive modulation Throughput Enhancement Scheme the system throughput increases promisingly as shown in Figure Figure 3.43 Comparative plot for system throughput enhancement scheme Fixed modulation technique QPSK produces high throughput of 38Mbps. Throughput of 120Mbps and 160Mbps was achieved in 16QAM and 64QAM. QPSK gets this response after 13dB of SNR and is capable of working till 25dB of SNR. Similarly, 16QAM gets constant maximum throughput after 29dB of SNR and is capable of upto 33dB. Compared to the previous two modulations, 64 QAM gets the extreme throughput of 160Mbps after 33dB of SNR with the help of proposed detection algorithm for interference cancellation and the signal detection.

51 106 In practice, high throughput at low SNR is necessary. This concept is appreciably got by Resource allocation method. Throughput is gradually increased when compared to the previous fixed modulation techniques. Again using proposed Enhancement Scheme, it is appreciably increased to a very high value. At 9dB of SNR, QPSK gets the throughput of 10Mbps and the proposed Resource Allocation scheme and the proposed throughput scheme achieved the improved throughputs of 25Mbps and 40Mbps. After 13dB of SNR, QPSK reach the maximum throughput of 35Mbps. Similarly, after 25dB of SNR, 16QAM achieved the maximum throughput of 120Mbps. 64QAM and the proposed resource allocation get its maximum value of throughput 160Mbps after 33dB of SNR. Depending on the number of subcarriers, the number of transmitter and receiver antennas, interference cancellation with detection technique, Adaptive modulation and improvement in the throughput enhancement by CSI technique selection, WLAN gets the high throughput of 180Mbps. This increase in data rate needs the Synchronization technique between transmitter and receiver Proposed Synchronization Estimator The Conventional OFDM synchronization can be divided into dataaided and non-data-aided categories. The data-aided category uses a training sequence or pilot symbols for estimation. It has high accuracy, low computational complexity, reduction in bandwidth and reduces the data transmission speed. The non-data aided category often uses the cyclic prefix correlation. Here bandwidth is not wasted but results in reduction of the transmission speed. Its estimation range is too small and not suitable for acquisition. Out of these two, data-aided category is suited for IEEE a WLAN standard. Two different approaches of synchronizations are used, one is time domain approach which is performed in received signal before it goes

52 through the FFT block and the other one is frequency domain approach which is performed after the FFT operation. 107 In OFDM-MIMO system two aspects of synchronizations are considered as symbol timing synchronization and carrier frequency synchronization. For the symbol timing recovery, a conventional zerocrossing method is not applicable, because the symbol has zero crossing points within the symbol period. Therefore, the symbol timing synchronization is difficult to carry out. Synchronization of an OFDM signal requires finding the symbol timing and carrier frequency offset. Symbol timing for an OFDM signal is significantly different from a single carrier signal since there is no eye opening where a best sampling time can be found. Rather there are hundreds or thousands of samples per OFDM symbol since the number of samples necessary is proportional to the number of subcarriers. Finding the symbol timing for OFDM means finding an estimate of where the symbol starts. There is usually some tolerance for symbol timing errors when a cyclic prefix is used to extend the symbol. Synchronization has been one of the crucial research topics in OFDM-MIMO system because of its sensitivity to the timing and frequency errors. To guarantee the fast and accurate data transmission, the Inter Symbol Interference (ISI) and Inter Carrier Interference (ICI) caused in the transmission have to be eliminated as much as possible. In OFDM system, ISI can be avoided by inserting cyclic prefix with length greater than the channel impulse response, and the ICI can be eliminated by maintaining the orthogonality of carriers under the condition that the transmitter and the receiver have exactly the same carrier frequency. But in the real world, frequency offsets will be arising from the frequency mismatch of the transmitter and the receiver oscillators and the existence of Doppler shift in the channel. In addition, due to the delay of signal during transmission in the