21-07-2012, 10:13 AM
MULTI USER DETECTION IN CDMA USING MATLAB
MULTI USER DETECTION IN CDMA USING MATLAB.doc (Size: 1.62 MB / Downloads: 77)
1. INTRODUCTION
1.1 Background
In most systems for mobile communication today, the available frequency spectrum is shared between different users by using separate time slots and/or frequency bands. In the third generation UMTS standard, the users are allocated different codes so that they can share the same frequency band – so called CDMA (Code Division Multiple Access).
This technique has several advantages; among other it gives a significant increase in robustness towards interference.
1.2 Objective
The goal of this project is to implement a real-time digital acoustic communication system using CDMA. The system should work reliably in a multi-path environment. The system should allow for different coding gain to illustrate the trade-off between data rate and robustness to interference.
1.3 Problem Specification
The implementation of a digital CDMA communications system using a transmitting loudspeaker connected to a DSP on one PC, and a receiving microphone connected to a DSP card on a second PC should be performed.
The system should be able to transmit data in real time. A number of different coding rates (the length of the spreading code) should be available. The system should work in the presence of narrowband signals (sinus tones). The system should also handle synchronization and possible frequency offset between the transmitter and receiver.
It should also provide a robust communication in the multi-path propagation provided by the acoustic channel. This may require channel equalization, i.e. the use of an equalizer in the receiver.
1.4 Modulation Methods
One of the major decisions that must be made in the construction of the communication system is to choose what modulation method to use. Since the only demand in the problem specification was that a construction of a robust CDMA communication system should be performed, a modulation method that is suited for this endeavor should be used.
In this section there will be a discussion of several methods that is possible to use. A short description of the pros and cons of each method can be seen below.
The decision of modulation method is largely built on extensive studies of different literature. Several books, old reports and publications have been used in an effort to choose a suitable method. Some experiments have also been performed, mainly in the channel estimation process.
1.4.1 Base band Modulations
The transmissions will occur over an acoustic channel. This kind of channel only passes signals within a band of frequencies that is removed from DC, i.e. a band pass channel. Therefore a base band modulation scheme, such as base band Pulse Amplitude Modulation (PAM) is not feasible.
1.4.2 Binary Frequency Shift Keying (BFSK)
There are two different frequencies used to transmit a BFSK modulated binary sequence. The two signal waveforms are
Studying these waveforms, one can clearly see that this is a non-linear modulation method, since the information is withheld in the frequencies.
The primary goal was to create a robust communication system. That includes having good Inter Symbol Interference (ISI) “rejection”. This in turn indicates that a use of an equalizer is recommended.
Since the BFSK modulation is a non-linear modulation, a construction of a minimum mean squared error (MMSE) equalizer will be difficult. The MMSE equalizer is very good at nullifying the effect of ISI and is not so complex, therefore this is a kind of equalizer that we would want to use. Another major drawback of this modulation scheme is that in order to perform a good demodulation, the use of two Phase Locked Loops (PLL) might be necessary.
1.4.3 Phase Shift Keying (PSK)
The other major modulation methods that have to be considered are the different kind of PSK modulation schemes. PSK modulation methods are well suited for spread spectrum systems and the special nature of a DS-CDMA communication system is compliant with a PSK .
1.4.3.1 Binary Phase Shift Keying (BPSK)
PSK uses M different phases to represent information in a transmitted sequence. In binary PSK the M is 2, i.e. there are only two different phases used. This results in two different output signals
Where fc is the carrier frequency, is phase that represent a symbol and gT(t) is the transmitting filter pulse shape, in our case a raised-cosine.
A choice of = 0 to represent a binary 0 can be made, and to use = to represent a binary 1. This can also be represented geometrically in a two-dimensional vector space. This space is spanned by two basis functions and . The resulting signal constellation can be viewed in figure 1.1.
After the carrier modulation the receiver receives a band pass signal that can be expressed as
r(t) = um(t) + n(t)
In BPSK the received signal will only be correlated with the basis function . The resulting output of the correlator is a noise corrupted signal component
where
The detector that makes the final decision then processes this signal component. Note that because of the special nature of our DS-CDMA scheme we have mapped the zeros and ones, i.e. 0 => -1 and 1 => 1. One of the problems with the BPSK receiver is that the local oscillators in the transmitter and receiver are not phase locked. This can result in a carrier-phase offset, in the received signal, which can be expressed as
This phase offset can be conquered in two ways, a PLL can be constructed or it can be removed by the use of an equalizer and a good synchronization process.
A PLL is quite difficult to construct and implement and this, together with the fact that a good equalizer is to be constructed for a better robustness in the system, the decision to exclude a PLL was made.
1.4.3.2 Quadrature Shift Keying (QPSK)
QPSK resembles in many ways BPSK, the information is represented by phases but unlike in BPSK there are four different phases in use. This means that a better use of the bandwidth can be achieved.
On the downside a QPSK means that the system will be more complex and in turn more difficult to implement. One of the goals is to have a high bit rate as possible, but this is not a specified demand. The main concern is the robustness and stability of the system.
This together with the fact that a BPSK system has been used in numerous previous projects with great result the decision not to use QPSK was made.
1.4.3.3 Minimum Shift Keying (MSK)
MSK is a modulation scheme that is even more difficult to implement than QPSK. Therefore this modulation method is not more suited for our communication system than QPSK is. And that of course means that the use of BPSK rather than MSK is preferred.
1.4.4 Channel Estimation
In order to decide which modulation scheme to use, and to get an idea of the limitations for the bit rate and a suitable carrier frequency, an estimation of the channel properties was performed.
In the communication system the channel consists of several different parts including the “real” acoustic channel. Other parts are D/A- and A/D-converters, the loudspeaker and the microphone. A discrete model of the channel can be viewed in figure 1.2. As can be seen in figure 1.2 All the different impulse responses of the channel, i.e. hl(t), ha(t) and hm(t) can be combined into one single impulse response called h(n). Where hl(t) means the loudspeaker, ha(t) the acoustic channel and hm(t) means the microphone.
The usable frequency span will be limited by the quality of the loudspeaker, the microphone and the DSP-cards. The acoustical environment will also affect the transmission in form of distortion due to reflections and noise.
A first step to get some knowledge about the channel properties is to send and record white noise over the channel. A frequency analysis of the recorded data gives an idee of a usable frequency span. Furthermore, a frequency analysis of a recording of the noisy environment without any signal, gives us an idee of were the disturbance is minimum.
In a normal acoustic environment the absorption of high frequencies are greater than for lower ones, this implies less interference due to reflections of high frequencies.
Figure 1.3 above shows a recording of the noise environment, i.e. the power spectrum of the channel without any transmission being performed. This spectrum is done by simply recording the acoustic channel without transmitting anything with the loudspeaker. It can clearly be seen in the figure that there is more noise in the lower part of the frequency spectrum. This means that with a lower carrier frequency the transmission will be more affected by noise, i.e. the bit error rate will rise. This in turn would indicate that a low frequency is not recommended
In the figure 1.4 above, the power spectrum of the acoustic channel when white noise is transmitted is depicted. This is done by letting the microphone record equally probable ones and zeros that are transmitted by a loudspeaker 40 cm away.
The attenuation of the acoustic channel is an important factor to take into account when deciding which carrier frequency to use. A high attenuation will mean that the signal strength of the transmission will be significantly reduced. This in turn will have the affect that the transmission will be more sensitive to noise and other interference.
From the figure we can see that the attenuation of the signal is fairly large above 10 kHz and quite unstable before 5 kHz. Taking into account both factors the conclusion must be that a carrier frequency between 5 and 10 kHz is recommended. All of this has resulted in that a decision of using a carrier frequency of 5 kHz was done.