 Research Article
 Open Access
 Published:
Blind Separation of Two Users Based on User Delays and Optimal PulseShape Design
EURASIP Journal on Wireless Communications and Networking volume 2010, Article number: 939340 (2010)
Abstract
A wireless network is considered, in which two spatially distributed users transmit narrowband signals simultaneously over the same channel using the same power. User separation is achieved by oversampling the received signal and formulating a virtual multipleinput multipleoutput (MIMO) system based on the resulting polyphase components. Because of oversampling, high correlations can occur between the columns of the virtual MIMO system matrix which can be detrimental to user separation. A novel pulseshape waveform design is proposed that results in low correlation between the columns of the system matrix, while it exploits all available bandwidth as dictated by a spectral mask. It is also shown that the use of successive interference cancelation in combination with blind source separation further improves the separation performance.
1. Introduction
We consider the problem of multiuser separation in wireless networks via approaches that do not use scheduling. This problem is of interest, for example, when traffic is generated in a bursty fashion, in which case fixed bandwidth allocation would result in poor bandwidth utilization. Lack of scheduling results in collisions, that is, users overlapping in time and/or frequency. To separate the colliding users, one could enable multiuser separation via receive antenna diversity, or code diversity, as in codedivision multipleaccess (CDMA) systems. However, the former requires expensive hardware since multiple transceiver front ends involve significant cost. Further, the use of multiple antennas might not be possible on smallsize terminals or devices. CDMA systems require bandwidth expansion, which requires greater spectral resources, and also introduces frequencyselective fading. In the following, we narrow our field of interest to randomaccess systems that for the aforementioned reasons cannot exploit antenna diversity, and that are inexpensive in terms of bandwidth. In such systems, the use of different power levels by the users can enable user separation by exploiting the capture effect [1], or successive interference cancellation (SIC) [2]. Different power levels can result from different distances between the users and the destination, or could be intentionally assigned to users in order to facilitate user separation. While the former case, when it arises, makes the separation problem much easier, the latter approach might not be efficient, as lowpower users suffer from noise and channel effects. In the following, we focus on the most difficult scenario of separating a collision of equalpower users. Almost equal powers would also result from power control. Power control is widely used, hence this scenario is of practical interest.
A delaydivision multiple access approach was proposed in [3], which exploits the random delays introduced by transmitters. The approach of [3] considers transmissions of isolated frames. It requires that users have distinct delays, assumes full channel knowledge at the receiver and exploits the edges of a frame over which users do not overlap. Pulseshape waveform diversity was considered in [4] to separate multiple users in a blind fashion. In [4], the received signal is oversampled and its polyphase components are viewed as independent mixtures of the user signals. User separation is achieved by solving a blind source separation problem. Although no specifics on waveform design are given in [4], the examples used in the simulations of [4] consider wideband waveforms for the users. However, if large bandwidth is available, then CDMA would probably be a better alternative to blind source separation. Pulseshape diversity is also employed in [5, 6], addressing situations in which the pulseshape waveforms have bandwidth constraints.
In this paper we follow the oversampling approach of [4], with the following differences. First, we introduce an intentional halfsymbol delay between the two users. Second, both users use the same optimally designed pulseshape waveform. Third, we use successive interference cancelation in combination with blind source separation to further improve the separation performance.
The paper is organized as follows. In Section 2, we describe the problem formulation. The proposed blind method is presented in Section 3. The Pulseshape design is derived in Section 4. Simulation results validating the proposed method are presented in Section 5, while concluding remarks are given in Section 6.
Notation 1.
Bold capitals denote matrices. Bold lowercase symbols denote vectors. The superscript denotes transposition. The superscript denotes the pseudoinverse. denotes the diagonal matrix with diagonal elements the elements of . denotes rounding down to the nearest integer. denotes the trace of its argument. denotes the phase of its argument.
2. Problem Formulation
We consider a distributed antenna system, in which users transmit simultaneously to a base station. Although much of this paper studies the case , for reasons that will be explained later, we will keep the user notation throughout. Narrowband transmission is assumed here, in which the channel between any user and the base station undergoes flat fading. In addition, quasistatic fading is assumed, that is, the channel gains remain fixed during several symbols.
The transmitted signal of user is of the form
where is the th symbol of user ; is the symbol period; is a pulseshaping function with support , where is an integer.
The continuoustime baseband received signal can be expressed as
where denotes the complex channel gain between the th user and the base station; denotes the delay of the th user; is the carrier frequency offset (CFO) of the th user, arising due to relative motion or oscillator mismatch between receive and transmitter oscillators, and represents noise.
Our objective is to obtain an estimate of each user sequence, , up to a complex scalar multiple that is independent of . The estimation will be based on the received signal only, while channel gains, CFOs and user delays are assumed to be unknown. During the recovery process, there is permutation ambiguity, that is, the order of the users may be lost and again the user signals will be recovered up to a scalar multiple. However, these are considered to be trivial ambiguities and are inherent in blind estimation problems.
We should note that typically, in highspeed communication systems, the main lobes of the pulseshape functions overlap by [7]. This extended time support allows for better frequency concentration, or equivalently, lower spectral occupancy for the transmission of each symbol. However, it introduces intersymbol interference (ISI). Examining for (see (1)), we note the contribution of the th symbol, the contribution of symbol due to the main lobe of , and also contributions of symbols due to the sidelobes of , respectively. If is a Nyquist pulse and samples are taken at times , the overlap does not play any role. However, when we obtain more than one sample during the symbol interval, we expect ISI effects.
Sampling the received signal at times we obtain
where is the normalized CFO between the th user and the base station, "*" denotes convolution, and is defined as
The th polyphase component, , can be expressed as
Let us form the vector as . It holds that
where is a matrix whose th row equals ; ; and . This is a instantaneous multipleinput multipleoutput (MIMO) problem. Under certain assumptions, to be provided in the following section, the channel matrix is identifiable, and the vector can be recovered up to certain ambiguities. In particular, for each , we get different versions of , that is, within a scalar ambiguity. The effects of the CFO on the separated signals can be mitigated by using any of the existing singleCFO estimation techniques (e.g., [8–13]), or a simple phaselocked loop (PLL) device [14].
3. Blind User Separation
3.1. Assumptions
The following assumptions are sufficient for user separation.

(A1)
Each of the elements of , as a function of , is a zeromean, complex Gaussian stationary random process with variance , and is independent of the inputs.

(A2)
For each , is independent and identically distributed (i.i.d.) with zero mean and nonzero kurtosis, that is, The 's are mutually independent, and each user has unit transmission power.

(A3)
The oversampling factor satisfies .

(A4)
The channel coefficients are nonzero.

(A5)
The user delays, , in (3) are randomly distributed in the interval .

(A6)
Either the CFOs are distinct, or the user delays are distinct.

(A7)
for ; and only for and .
Under assumption , it is easy to verify that the rotated input signals are also i.i.d. with zero mean and nonzero kurtosis. Also, the 's are mutually independent for different 's. Assumptions and are needed for blind MIMO estimation based on (7). Assumptions guarantee that the virtual MIMO channel matrix in (7) has full rank with high probability. Assumption can actually be relaxed. As will be discussed later, (see (18)), the contributions of lowvalue columns of in (7) can be viewed as noise. This effectively reduces the dimensionality of the problem. and guarantee that will be nonzero for all allowable values of , , and . To see the effect of , let us write the channel matrix as
where is formed by appending in (5) for different 's, that is,
and consider the case in which all users have the same delays, that is, . If the CFOs are different, A has full column rank. Even if the CFOs are not distinct, the columns of the channel matrix can be viewed as having been drawn independently from an absolutely continuous distribution, and thus the channel matrix has full rank with probability one [15].
3.2. Channel Estimation and User Separation
One can apply to (7) any blind source separation algorithm (e.g., [16]) to obtain an estimate of the channel matrix, , which is related to the true matrix as
where is a column permutation matrix, and is a complex diagonal matrix. The method of [16] requires fourthorder cumulants of . Accordingly, the estimate of the decoupled signals within permutation and diagonal complex scalar ambiguities is
Denoting by the diagonal element of , which corresponds to the phase ambiguity of user with delay , the separated signals can be expressed as
At this point, the users' signals have been decoupled, and all that is left is to mitigate the CFO in each recovered signal. This can be achieved with any of the existing single CFO estimation methods, such as [8–12], or [13]. Alternatively, if the CFO is very small, then we can estimate it and at the same time mitigate its effects using a PLL. We should note here that even a very small CFO needs to be mitigated in order to have good symbol recovery. For example, for 4ary quadrature amplitude modulation (4QAM) signals and without CFO compensation, even if the normalized CFO is as small as , the constellation will be rotated to a wrong position after samples.
If the CFO is large, then a PLL does not suffice, with the severity of the problem depending on the modulation scheme. In this case, the phase of the estimated channel matrix can be used to obtain a CFO estimate. If for all , then it can be easily seen that with
where is a vector with all elements equal to one, and . The leastsquares estimates of the CFO can be obtained as
where is the th element of .
On noting that the decoupled signals in (12) are permuted (see (11)) in the same manner as the estimated CFOs in (14), we can use the 's to compensate for the effect of CFO in the decoupled signals in (12) and obtain estimates of the input signals as
where with . In order to resolve user permutation and shift ambiguities, one can use user IDs embedded in the data [17].
Although in theory, under the above stated conditions, the matrix has full rank for any number of users, , the matrix condition number may become too high when CFOs or delay differences between users become small. As increases, the latter problem will escalate. Further, for large , the oversampling factor, , must be large. However, as increases, neighboring pulseshape function samples will be close to each other, and the condition number of will increase. Therefore, the shape of the pulseshape function sets a limit on the oversampling factor one can use and thus on the number of users one can separate. Recognizing that the above are difficult issues to deal with, we next focus on the twouser case. Further, we propose to introduce an intentional delay of between the two users, in addition to any small random delays there exist in the system.
The performance of user separation depends on the pulseshape function and also on the location of the samples. Although uniform sampling was described above, nonuniform sampling can also be used, in which case the expressions would require some straightforward modifications. If the samples correspond to a lowvalue region of the pulse, the corresponding polyphase components will suffer from low signaltonoise ratio. Also, if the sampling points are close to each other, then the condition number of will increase. Therefore, one should select the sampling points so that the corresponding samples are all above some threshold and the sampling points are as separated as possible. The effect of pulseshape and optimal shape design will be discussed in the following section.
4. PulseShape Design
In this section, we first investigate the effects of pulseshape on the condition number of . Since the condition number of a matrix increases as the column correlation increases, we next look at the correlation between the columns of .
Let us partition the channel matrix into two submatrices and , containing, respectively, the columns of corresponding to the main lobe and those corresponding to the sidelobes of the pulse. We can rewrite (7) as follows:
where
with as defined in (9). Correspondingly, , and . If the sidelobes of the pulse are very low, then can be treated as noise and (16) can be written as
4.1. Pulse Effects
In order to maintain a wellconditioned , the correlation coefficient between its columns should be low. Let us further divide the matrix into and . The elements of are samples from the decreasing part of the main lobe of the pulse. On the other hand, the elements of are from the increasing part of the main lobe of the pulse. Thus, the correlation coefficient of and is smaller than the correlation coefficient of and , or that of and . Thus, we focus on the effects of the pulse on the column correlations within and .
Proposition 1.
Let be a Nyquist pulse that is positive within its main lobe, that is, for . We further assume is an even function with very low sidelobes. For and () in , the absolute value of the correlation coefficient between and is upper bounded as follows:
where , is the sampling interval, that is, , and is given by
where denotes the firstorder derivative of .
Proof.
See the appendix.
When is large, the following approximation holds:
Thus, for fixed and , the correlation coefficient between and decreases with increasing . It can be shown that the same holds for the correlation coefficient between and .
Because should be a Nyquist pulse with small sidelobes and for , it should hold that
where is small.
There are additional constraints that the pulse should satisfy, the most important of which is a bandwidth constraint. Most commercial systems, for example, the IEEE 802.11a, IEEE 802.11b, and IEEE 802.11g wireless localarea networks (WLANs) [18], are equipped with a spectral mask that dictates the maximum allowable spectrum, or equivalently, the maximum symbol rate. This leads to a constraint of the form
where is the Fourier transform of , and denotes the spectral mask.
4.2. Optimum Pulse Design
Based on the above constraints and assuming that satisfies the conditions of Proposition 1, the pulse design problem can be expressed as
The problem of (24a)(24e) is not easy to solve. Next, we will take steps towards reformulating it into a convex optimization problem. Let be a vector containing samples of taken in , with sampling interval , in which case ( is an integer representing the number of samples in each symbol interval). The objective function (24a) is equivalent to
where is of the form
As is an even symmetric function, the Fourier transform of can be represented as , where , with power spectral density (PSD) equal to . Hence, the constraint (24b) is equivalent to
Because (27) involves an infinite number of constraints, we sample in the frequency domain:
where is the number of samples in . In order for (28) to be a good approximation of (27), should be on the order of [19].
In the discretetime domain, (24c) is equivalent to
where is small and with leading zeros.
Define , with the th element equal to . Equation (24d) is equivalent to
with . Hence the problem of (24a)(24e) can be reformulated as
Since it involves maximization of a convex function, (31a) is not a convex optimization problem. Letting , should be a positive semidefinite matrix of rank . The problem of (31a)–(31e) is equivalent to
However, the constraint of (32g) is not a convex constraint. By dropping it, we obtain a semidefinite relaxation of the primal problem [20]. The resulting convex optimization problem is
As we drop the constraint , the resulting might not be of unit rank. In this case, we apply eigendecomposition to . Let
where is the largest eigenvalue of , and is the corresponding eigenvector. As , its eigenvalues for . If
then can result in a good pulseshape. If , then it holds that
which indicates that in the problem of (24a)–(24e) is maximized. Moreover, can guarantee the validity of (31b) and (31c). Also, if and , then
This indicates that the PSD of will be under the IEEE 802.11 mask. In the same way, we can prove that
which further indicates that has small sidelobes. Moreover, is small and the validity of (38) implies that
which indicates that, if we sample at intervals , the interference from neighboring symbols can be neglected.
If , then it holds that . Also, (33e) requires that the th element of be greater than zero for . Hence, or for . Thus, within its mainlobe, is greater than zero, or its amplitude becomes very small.
5. Simulation Results
5.1. Pulse Design Examples
In this section, we demonstrate the performance of a pulse designed as described in Section 4.2. We take samples per symbol, that is, , and set . Then we obtain and samples in the time and frequency domains, respectively. We take to be . In Figure 1, we show the ratio of the resulting matrix at different symbol rates, where is the largest eigenvalue of . One can see that the smallest is above , which means that the condition of (35) is satisfied. Therefore, is a good choice of pulseshape.
For symbol rate , or equivalently, , the designed timedomain pulse is shown in Figure 2. For comparison, the Isotropic Orthogonal Transform Algorithm (IOTA) pulse [21] is also shown in the same figure. The corresponding PSDs, along with the IEEE 802.11 spectral mask are given in Figure 3. From the figures we can see that the proposed pulse decreases faster than the IOTA pulse within . The larger the value of , the faster decreases. In Figure 3, one can see that the PSD of the proposed pulse is under the 802.11 mask, while the PSD of the IOTA pulse violates the mask at MHz.
For symbol rate , or, , the obtained pulse is given in Figures 4 and 5. We also plot the raised cosine pulse with rolloff factor being equal to . One can see that, in the frequency domain, the proposed pulse is under the 802.11 mask, while in the time domain the proposed pulse is narrower. Note that at this symbol rate, the IOTA pulse cannot meet the mask constraint.
5.2. SER Performance
In this section, we demonstrate the performance of the proposed user separation approach via simulations. We consider a twouser system. The channel coefficients and are taken to be zeromean complex with unit amplitude and phase that is randomly distributed in . The CFOs are chosen randomly in the range . The input signals are QAM containing symbols. The estimation results are averaged over independent channels, and MonteCarlo runs for each channel. One user is intentionally delayed by half a symbol and in addition, small delays, taken randomly from the interval , are introduced to each user.
In our simulations, we combine blind source separation method with SIC [2]. For blind source separation the Joint Approximate Diagonalization of Eigenmatrices (JADE) algorithm was used, which was downloaded from http://perso.telecomparistech.fr/~cardoso/Algo/Jade/jade.m. We first apply JADE to decouple the users, and then correct the decoupled users' CFOs. Subsequently, the strongest user, that is, the one which shows the best concentration around the nominal constellation is deflated from the received polyphase components to detect the other user. SIC requires that the first user should be detected very well. To achieve this, the sampling points are chosen around the peak of one user signal, so that ISI and interuser interference effects are minimized.
Eliminating CFO effects from the decoupled users can be done via a PLL, if the CFO is small, or a PLL initialized with a good CFO estimate, if the CFO is large as the PLL by itself would not converge in this case. For the latter case, since we sample around the peak of one user, the CFO estimation formula of (14) requires a small modification before it is applied. Let the sampling points occur at , and let be the phase of the channel matrix corresponding to these sampling points. The leastsquares estimate of the CFO can be obtained as
where is the th element of .
In this experiment, the pulse has time support . We take polyphase components of the received symbols, each consisting of samples taken evenly over the interval , with sampling period . In order to sample around the peak of one user, we used the true shift values. However, in a realistic scenario this information would be obtained via synchronization pilots [17].
The symbol error rate (SER) performance at sec, that is, at symbol rate M/sec, using the waveform of Figure 2, is shown in Figure 6 along with the performance corresponding to the IOTA pulse. We can see that the SER of the proposed pulse is lower; there is an approximate dB SNR advantage over the IOTA result.
In Figure 7, we show the SER versus SNR at different symbol rates. First, by taking sec, or equivalently symbol rate M/sec, we compare the SER performance of the proposed pulses and the raised cosine pulse with rolloff factor 1. As we can see, the performance of the proposed pulse is better. For example, the proposed pulse can achieve at dB SNR, while the raised cosine pulse needs dB SNR to achieve the same SER. In the same figure we show the SER performance of the proposed pulse at symbol rate M/sec. At this rate, the proposed pulse can achieve an SER of at dB SNR.
In Figure 8, we show SER performance for different values of the oversampling factor, , at different symbol rates. For , the sampling occurs evenly within the interval of each received symbol with sampling period . One can see that, for symbol rate M/sec, when the SNR is higher than dB, the SER performance improves by increasing from to . For symbol rates equal to M/sec and M/sec the SER performance remains almost the same with increasing .
In order to demonstrate the effect of the proposed pulse on the condition number of the system matrix, we show in Figure 9 the condition number of corresponding to the proposed and IOTA pulses, averaged over random channels realizations and with . In order to make a fair comparison, the CFOs and random delays were set to be the same for both pulses. No noise was added in the data. The estimated 's were collected from the JADE output, and their condition numbers were calculated. One can see that the proposed pulse results consistently in lower condition number than the IOTA pulse.
Next, we show the effect of user delays on performance. As before, one user is delayed by a halfsymbol interval, and in addition, a random delay is added to both users to model random delays introduced at the transmitter. In this experiment, the range for the random delay is increased from to . For random delays within , in order to prevent the delay difference of two users from being too small, we select the delays so that their difference is no less than a threshold . The resulting SER performance is shown in Figure 10. When the range of increases from to the performance becomes worse. This is because by increasing the range for the random delay, the signals of the two users overlap by a larger amount, which results in high condition number for the channel matrix . The best performance would be obtained with just the halfsymbol delay and no random delays; however, this is not a realistic case.
Next, to show the advantage of the intentional halfsymbol delay, we consider a case without intentional delay, with random user delays only. The random delays of both users are taken within . In order to prevent worsening of performance we restricted the smallest delay difference between two users to be no less than . We compare the SER performance of the proposed pulse with IOTA and raised cosine pulses at different symbol rates. Firstly, comparing the corresponding curves in Figure 10, one can first see that without the intentional delay the SER performance decreases. In particular, for the proposed pulse in order to achieve SER , we need an SNR of dB and dB for symbol rates M/sec and M/sec, respectively. Secondly, the SER performance of the proposed pulse is still better than that of IOTA and raised cosine pulses at the corresponding symbol rate.
Finally, we show the effect of CFOs on performance (see Figure 12). In order to highlight the effect of the CFOs, SER results were obtained without intentional delay, with random delays taken in the interval and by setting the delay difference of the two users to be no less than . The normalized CFOs were chosen randomly within the range for , and . For we restricted the smallest difference between two CFOs to be no less than , and for , we set no threshold on the CFO difference of the two users. For , the CFO is quite large, and the PLL by itself is not enough to remove the CFO in the decoupled users. Therefore, we first used the method described in Section 3.2 to estimate the CFOs and then used the PLL to compensate for the residual CFO.
The quality of the CFO estimates depends on the accuracy of the channel matrix estimate. Since lowmagnitude elements of the channel matrix correspond to low values of the pulse, and as such are susceptible to errors, we set a threshold, , defined as , and for CFO estimation, we only use elements of whose amplitudes are greater than . In this experiment, we took . The CFO effects were eliminated via a PLL initialized with the CFO estimate of (40). One can see that the larger gives better performance. It is important to note that the large CFOs involve bandwidth expansion. The percentage of bandwidth expansion can be calculated as , where MHz is the bandwidth of the pulse. For and , the percentages of bandwidth expansion for symbol rates M/sec, M/sec and M/sec are, respectively, , , and .
6. Conclusions
A blind user separation scheme has been proposed that relies on intentional user delays, optimal pulseshape waveform design, and also combines blind user separation with SIC. The proposed approach achieves low SER at a reasonable SNR level. Simulation results for the case have confirmed that the proposed pulse design leads to SER performance better than that of conventional pulseshape waveforms. The intentional delay was equal to half a symbol interval, which means that the users still overlap significantly during their transmissions. The use of intentional delay is necessitated by the fact that, although small user delay and CFO differences help preserve the identifiability of the problem, in practice, they may not suffice to separate the users. Also, although the proposed approach can work for any number of users, as the number of users increases, the CFO and delay differences become smaller, which makes the separation more difficult. Based on our experiments, small CFO differences did not affect performance. Although introducing large intentional CFO differences among users could help, that would increase the effective bandwidth. A new ALOHAtype protocol that separates secondorder collision based on the ideas described in this paper, along with a softwaredefined radio implementation can be found in [17].
References
 1.
Onozato Y, Liu J, Noguchi S: Stability of a slotted ALOHA system with capture effect. IEEE Transactions on Vehicular Technology 1989, 38(1):3136. 10.1109/25.31132
 2.
Tse D, Viswanath P: Fundamentals of Wireless Communication. Cambridge University Press, Cambridge, UK; 2005.
 3.
BrandtPearce M: Signal separation using fractional sampling in multiuser communications. IEEE Transactions on Communications 2000, 48(2):242251. 10.1109/26.823557
 4.
Zhang Y, Kassam SA: Blind separation and equalization using fractional sampling of digital communications signals. Signal Processing 2001, 81(12):25912608. 10.1016/S01651684(01)001554
 5.
Petropulu AP, Olivieri M, Yu Y, Dong L, Lackpour A: Pulseshaping for blind multiuser separation in distributed MISO configurations. Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP '08), MarchApril 2008, Las Vegas, Nev, USA 27412744.
 6.
Liu X, Oymak S, Petropulu AP, Dandekar KR: Collision resolution based on pulse shape diversity. Proceedings of the IEEE International Workshop on Signal Processing Advances for Wireless Communications (SPAWC '09), June 2009, Perugia, Italy 409413.
 7.
Sklar B: Digital Communications: Fundamentals and Applicatons. Prentice Hall, Upper Saddle River, NJ, USA; 2001.
 8.
Ciblat P, Loubaton P, Serpedin E, Giannakis GB: Performance analysis of blind carrier frequency offset estimators for noncircular transmissions through frequencyselective channels. IEEE Transactions on Signal Processing 2002, 50(1):130140. 10.1109/78.972489
 9.
Ghogho M, Swami A, Durrani T: On blind carrier recovery in timeselective fading channels. Proceedings of the 33rd Asilomar Conference on Signals, Systems, and Computers, 1999, Pacific Grove, Calif, USA 1: 243247.
 10.
Gini F, Giannakis GB: Frequency offset and symbol timing recovery in flatfading channels: a cyclostationary approach. IEEE Transactions on Communications 1998, 46(3):400411. 10.1109/26.662646
 11.
Scott KE, Olasz EB: Simultaneous clock phase and frequency offset estimation. IEEE Transactions on Communications 1995, 43(7):22632270. 10.1109/26.392969
 12.
Wang Y, Ciblat P, Serpedin E, Loubaton P: Performance analysis of a class of nondataaided frequency offset and symbol timing estimators for flatfading channels. IEEE Transactions on Signal Processing 2002, 50(9):22952305. 10.1109/TSP.2002.801919
 13.
Roman T, Visuri S, Koivunen V: Blind frequency synchronization in OFDM via diagonality criterion. IEEE Transactions on Signal Processing 2006, 54(8):31253135.
 14.
Wolaver DH: PhaseLocked Loop Circuit Design. Prentice Hall, Englewood Cliffs, NJ, USA; 1991.
 15.
Sidiropoulos ND, Giannakis GB, Bro R: Blind PARAFAC receivers for DSCDMA systems. IEEE Transactions on Signal Processing 2000, 48(3):810823. 10.1109/78.824675
 16.
Cardoso JF, Souloumiac A: Blind beamforming for nonGaussian signals. IEE Proceedings F 1993, 140(6):362370.
 17.
Liu X, Kountouriotis J, Petropulu AP, Dandekar KR: ALOHA with collision resolution (ALOHACR): theory and software defined radio implementation. IEEE Transactions on Signal Processing 2010, 58(8):43964410.
 18.
IEEE standard 802.11 http://www.ieee802.org/11/
 19.
Wu S, Boyd S, Vandenberghe L: FIR filter design via spectral factorization and convex optimization. In Applied and Computational Control, Signals and Circuits. Edited by: Datta BN. Birkhauser, Basel, Switzerland; 1998:215245.
 20.
Sidiropoulos ND, Davidson TN, Luo ZQ: Transmit beamforming for physicallayer multicasting. IEEE Transactions on Signal Processing 2006, 54(6):22392251.
 21.
Le Floch B, Alard M, Berrou C: Coded orthogonal frequency division multiplex [TV broadcasting]. Proceedings of the IEEE 1995, 83(6):982996. 10.1109/5.387096
Acknowledgment
This research was been supported by the National Science Foundation under Grants CNS0916947 and CNS0905398 and by the Office of Naval Research under Grants N000140710500 and N000140910342.
Author information
Appendix
Proof of Proposition 1
Proof.
We have
On letting and , the correlation between and is equal to
Taking the absolute value of (A.2), we have
The last step is due to the fact that for within the mainlobe, and that both and are within the interval . We also have
When the number of samples is large, and are close. Applying firstorder Taylor series expansion to each component of , we have
On writing
we can approximate by
The correlation between and is bounded by
And can be approximated by
where . Since
we obtain
On substituting (A.11) into (A.9), we have
In the last step of (A.12), we assumed that is large and and are small, and also that .
In the same way can be approximated as
Because is an even function and also has very low sidelobes, (A.13) can be further simplified as
Substituting (A.12) and (A.13) into (A.8), we have
In this last step, we let
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
About this article
Received
Accepted
Published
DOI
Keywords
 Channel Matrix
 Blind Source Separation
 Successive Interference Cancellation
 Symbol Error Rate
 Symbol Rate