The trend towards larger bandwidths in mobile communications requires coarse quantization to obtain energy efficient analog-to-digital converters. The question is whether quantization and subsequent processing can be optimized with respect to relevant mutual information. Current standards rely on orthogonal frequency division multiplexing (OFD M) as air interface and require the implementation of an fast Fourier transform (FFT). Inspired by the efficient implementation of low density parity check (LDPC) decoders via lookup-tables optimized by the information bottleneck (IB) method, we suggest to approximate the computation of intermediate quantities inside the FFT with IB-optimized lookup-tables being designed offline. Consequently, the FFT only relies on the mapping quantization indices for which the flow of relevant information is optimized.

This paper considers a scenario being widely known as the Chief Executive Officer (CEO) problem under log-loss distortion. It consists of a set of distributed sensors observing statistically dependent data and forwarding their measurements to a common receiver. As these links are capacity limited, the sensors have to compress or quantize the data. This paper contributes an algorithm for the quantizer design based on the information bottleneck (IB) method. This algorithm successively designs scalar quantizers using the statistics of other quantizers as side-information. Furthermore, it allows individual compression rate adjustments in order to fulfill given rate-constraints of the capacity limited links. Numerical results demonstrate the benefit compared to individual scalar optimized quantization.

Multicasting popular contents (e.g. HD video streaming) with high data rates is a typical scenario in the future 5G system. In this paper, we investigate the downlink design of multicast Fog Radio Access Network (F-RAN), where each Remote Radio Head (RRH) is equipped with a cache module and has some limited signal processing functionalities. Caching some popular contents at the off-peak hours might not only greatly decrease the service delay, but also, as we are going to show, reduce the energy consumption of the network. Hence, the minimization of network energy consumption is considered within this work. In particular, information compression based (soft) fronthauling and information sharing based (hard) fronthauling are discussed. Based on only imperfect CSI, we propose algorithms for robust network design, where RRH deactivation, power and fronthauling traffic control, beamformer and precoder construction, etc., are jointly optimized for soft and hard fronthauling, respectively.

In this paper, we study the robust design of downlink multicast Cloud Radio Access Network (C-RAN), equipped with wireless backhaul. Due to feedback delays, channel estimation or quantization errors, etc., only impaired Channel State Information (CSI) might be available at the Central Processor (CP). In order to achieve a power-efficient system, we aim to minimize the network power cost while guarantee the Quality of Service (QoS) for all delivered contents, under several practical constraints of C-RAN. Despite the uncertainties introduced by the impaired CSI, we propose an algorithm, yielding a robust backhaul resource and power allocation mechanism. Moreover, the clustering formulation for Radio Units, as well as the robust multicast beamforming vectors, can also be derived. Numerical results demonstrate the effectiveness of such robust design in counteracting the channel uncertainties, and confirm the robustness in fulfilling the network constraints.

This paper is concerned with practical quantizer design for Compress-and-Forward (CF) in combination with Orthogonal Frequency Division Multiplexing (OFDM) in the 3-node relay channel. The Information Bottleneck (IB) method is an appropriate tool for this task as it maximizes relevant information given a specific restriction on the rate. Optimal quantization is not possible due the high complexity (dimensionality) of the problem since channel knowledge needs to be considered in addition compared to an Additive White Gaussian Noise (AWGN) channel. Therefore, recently introduced Information Bottleneck Graphs (IBGs) are used to simplify the problem. In this work, IBGs will be extended to exploit as well side information as stated by Wyner and Ziv. Furthermore, different IBGs will be distinguished to find the best trade-off between feasibility and exploitation of as much information as possible. Results in terms of achievable rates with a Bit Interleaved Coded Modulation (BICM) scheme reveal that the derived approaches lead to a better overall performance especially the more side information is considered at the relay.

The Radio Access Network based on Fog computing (F- RAN) has been recently proposed to alleviate the heavy burden on backhauls of Cloud Radio Access Network (C-RAN). Equipping the edge device of C-RAN with local storage module (cache) is a specific realization of F-RAN. Some frequently requested files are cached at the Radio Units (RUs), both the traffic load and the overall delay can be reduced. In this paper, we consider the downlink of the Cache-Enabled F-RAN, where minimization of power consumption is investigated for the Green Communication. We focus on the design of green network at the system level, such that not only the transmission power, but also all additional operational power of active RUs (incl. power used for circuits, associated backhaul, etc.) is considered. Based on the channel states, requested and cached files, an efficient algorithm for the joint optimization of RU selection, clustering formulation and beamforming is proposed. Meanwhile, the desired Quality of Service (QoS) is guaranteed and the load on each active backhaul is balanced according to its capacity. Numerical results show that with the proposed algorithm, more RUs can be switched off to achieve greener network, as the cache memory is increased.

The compression step of the Cloud Radio Access Networks (C-RAN) is studied in this paper. In order to maximize the achievable sum rate in the uplink of C-RAN, constrained by the finite-capacity fronthaul links to the Central Processor (CP) in the cloud, an optimization algorithm of quantizers used for compression is proposed. In C-RAN, there are multiple Radio Units (RUs) compressing and forwarding the correlated received signals simultaneously. Due to the correlations, Wyner-Ziv coding is used. Hence a joint optimization of the quantizers is required at the CP. In this paper we extend the Information Bottleneck (IB) method for this joint optimization. It is a combination of the Alternating Information Bottleneck method (AIB) and the Alternating Bi-Section method, which are both proposed in this paper. Then we use the proposed algorithms to study some key features of C-RAN.

Both scalar and vector quantization are studied when Compress and Forward (CF) is performed in the Two Way Relay Channels. The optimization schemes of these two kinds of quantizers aiming to maximize the sum rate are proposed. The optimization schemes are a combination of the Information Bottleneck (IB) method and the Bi- Section method. The performance of these two types of quantizers are compared. Then we compare the optimized CF with other relaying schemes, i.e., Decode and Forward (DF) and optimized Quantize and Forward (QF).

In this paper we consider the Two Way Relay Channel (TWRC) where Noisy Network Coding (NNC) is performed. We propose an optimization scheme in terms of the sum rate by combining the Information Bottleneck (IB) method and the Bi-Section method. Then we compare the performance of the quantizer optimized by the proposed algorithm with other quantizers. We also compare and analyse the maximized sum rate with the results when Compress and Forward (CF), Decode and Forward (DF) and Quantize and Forward (QF) are performed. The simulation results demonstrate the correctness and effectiveness of the proposed algorithm.

The application of non-orthogonal waveforms is one promising approach to meet the requirements of future mobile radio systems. The Gaussian waveform can considerably decrease the sensitivity to distortions and synchronization errors. The shape of the impulse can be adjusted according to the statistics of the channel, but the positions of lattice points in time-frequency-grid need to be adapted as well. It is shown that optimizing power allocations does lead to considerable improvements in the high Signal to Noise Ratio (SNR) region. To avoid the increased computational effort of the power allocation, the optimal lattice spacing is derived for Gaussian input alphabets as well as discrete modulation schemes. It is shown that adjusting the lattice spacing can achieve the same gain as a tedious power allocation. Furthermore, an analytical expression for a high SNR approximation of the optimal spacing is introduced.

Relaying techniques are a well investigated subject in literature. The simplest relay channel is a 3-node network and the communication is split into broadcast phase and Multiple Access (MAC) phase. The focus of this paper lies on the MAC part of the relay channel which is a superposition of the source's and the relay's signals. From information theory it is known that a trade-off between distributed Beamforming (dBF) and the classical MAC has to be found. Generally, the source transmits a superposition of a beamforming part and new information, where the power distribution between both parts has to be optimized. In this work, this conventional approach will be reinterpreted from a Multilevel Coding/Code (MLC) point of view, where the new information can be seen as a level on top of the information transmitted by the relay. In addition, Superposition Modulation (SPM) will be used as a joined mapping scheme to ensure coherent superposition of the dBF part. In this MLC view, the conventional scheme is extended to a more generalized scheme which allows to achieve the theoretical Decode and Forward (DF) rate of Gaussian relay channels. Furthermore, it is possible to realize a low complexity joint demapping of both signals at the destination with only minor performance degradation. Indeed, results show that the new scheme will outperform practical mapping schemes like Quadrature Amplitude Modulation (QAM).

Orthogonal Frequency Division Multiplexing (OFDM) is said to be extremely sensitive to synchronization offsets. Hence, non-orthogonal impulse shaping might be used in Generalized Frequency Division Multiplexing (GFDM) systems to increase robustness. In this paper GFDM with Gaussian pulse shaping is compared to a Cyclic Prefix (CP)-OFDM system where the receiver suffers from synchronization offsets in time and frequency domain. As the increased robustness due to non-orthogonal waveforms does not pay off, the GFDM system is clearly outperformed by the CP-OFDM system. Different results are obtained if a linear equalizer is added to the GFDM system. Even if only the interference caused by adjacent lattice points is treated by an equalizer, GFDM can be superior. Furthermore, we show that the performance of the non-orthogonal system can be adjusted dynamically to the channel conditions by changing the lattice spacing. Contrary to an orthogonal system this is possible without redesign of the transmit filters. The optimal lattice spacing follows from an interference/spectral efficiency trade-off that depends not only on the SNR but also on the modulation alphabet, where contradictory strategies are found for BPSK and 16QAM schemes.

This paper analyses the estimation process for Finite Rate of Innovation (FRI) signals. The main contribution is the derivation of the well known Cramér-Rao Bound (CRB) for the estimation of signal parameters for a pulse stream. Other publications consider the estimation of the signal instead of its parameters or omit the effect of sampling. In this contribution both effect are considered and analytical expressions for the Fisher Information Matrix are obtained. Furthermore, for the estimation of parameters of a single pulse analytical expressions for CRB are given and for the case of multiple pulses dependencies of the CRB of the distance between the pulses are illustrated.

The combination of Physical-Layer Network Coding (PLNC) and Orthogonal Frequency Division Multiple Access (OFDMA) in Two Way Relay (TWR) systems enables promising designs for future generation wireless systems. So far, existing resource allocation schemes for OFDMA in TWR systems do not exploit the advantage of PLNC, i.e. the relay is not necessarily interested in the original user messages, but their XOR combination. Recently, it was shown that the achievable throughput in a TWR system with PLNC is close to its mutual information. Thus, the mutual information might be well suited to address the power allocation problem of a PLNC system with OFDMA. However, as no closed form expression is known for the mutual information of such a system, the optimal resource allocation cannot be found easily. Thus, an appropriate approximation for the mutual information in the critical MAC phase is proposed within this paper. Afterwards, that approximation is used to optimise the power allocations in an OFDMA system with individual power constraints. Simulation results and a comparison to alternative low effort algorithms confirm the efficiency of this approach.

This paper proposes a novel system design for signal acquisition in Particle Tracking Velocimetry (PTV) applications. For exploiting the Finite Rate of Innovation (FRI) of these signals an appropriate sampling kernel is designed and the reconstruction process is illustrated. The main contribution is to show that aliasing effects can be neglected under certain conditions and an exact reconstruction is possible in most cases. While the sampling rate is still at least twice the FRI, it can be a fraction of the sampling kernel bandwidth. Hence, the sampling rate can be adapted to the physical process under investigation without changing the kernel designed for the worst case scenario. Considering practical aspects of the reconstruction process it is shown that even in the presence of noise precise results can be obtained.

This paper addresses the sampling of Finite Rate of Innovation (FRI) signals and proposes new lower bounds for accurate reconstruction. Recently, the FRI approach has been established next to the Compressed Sensing framework to deal with signals which are sparse in their parametric description. Many authors showed that a reconstruction is exactly possible in the absence of noise and, therefore, lower bounds on the sampling rate have been presented. Furthermore, different sampling kernels have been proposed such as the Sum of Sinc (SoS) kernel or the Gaussian low pass kernel. This contribution is motivated by a practical system design for sampling FRI signals with a fixed SoS filter design but varying Rate of Innovation (RoI). Therefore, a varying sampling rate is desired as well. For the proposed system design sampling below the previously presented bounds, i.e. below twice the kernel bandwidth, is analysed. This means sampling is applied in a sub-Nyquist regime with respect to the sampling kernel. The aliasing effect is analysed and new bounds are presented which allow a stable reconstruction. This analysis is supported by numerical simulation.

Within this paper we address the energy minimisation problem for a multi-hop relay networks where incremental redundancy (or mutual-information accumulation) is used as relaying strategy. The problem is to find optimal power levels and transmit rates, or equivalently optimal power and time slot lengths. First, we explain how the structure can be exploited to solve the problem with numerical optimisation methods. However, this approach is inefficient for larger networks as the number of optimisation parameter increases faster that the number of nodes in the network. Based on the results obtained by the numerical methods, we propose a very efficient resource allocation algorithm whose complexity increases only linearly with the number of nodes. Furthermore, the scheme can be implemented in a distributed way, since only partial channel state information is required for resource allocation at each node.

Recent research has developed a new, entirely digital architecture, called X-ORCA, that determines the phase shift of two periodic signals with a resolution as good as about 20 ps. This paper incorporates the X-ORCA system into a wireless experimental setup to form a localization system. The practical experiments utilize a 2.484 GHz transmitter and run the X-ORCA core on a Cyclone II FPGA. The results indicate that this simple localization system easily yields a spatial resolution in the sub-millimeter range.

This paper considers the resource allocation problem for a Multiple Access Relay Channel (MARC). A transmission scheme with OFDMA and Amplify-and-Forward (AF) relaying is assumed. To maximise the sum rate a joint allocation of transmit power and subcarriers has to be done. This problem is known to be NP-hard and can only be solved by an exhaustive search which is not feasible for practical problem sizes. Different near-optimum algorithms for solving this problem are proposed. The problem of carrier and power allocation can either be done separately or jointly. For example a straight forward gradient method can be applied if the constraints of the allocation problem are relaxed. The obtained solution tends to fulfil the original constraints and it is also shown to be nearly optimum. The performance of all algorithms is evaluated by numerical simulations. Furthermore, their complexities are analysed and compared.

In this paper, we provide a semi-analytical method for predicting the upper and lower bound of the end-to-end mutual information for a relay network with practical codes and different combining and decoding mechanisms. Especially in the case of non error-free decoding at the relays, the prediction can be used to significantly improve the performance of the system under consideration.

Adaptive relaying schemes significantly improve the performance of relay networks by switching the forwarding technique according to the reliability of the received signal. In this paper, semi-analytical solution for the power optimization in adaptive relay network is derived. Furthermore, the accuracy of the solution is confirmed by using the polyblock algorithm, which is an optimization tool for monotonic programs. The results show that significant power gains can be achieved by the devised algorithm.

Analyzing the capacity of relay networks is often based on the max-flow min-cut theorem providing an upper bound on the true capacity which is generally still unknown. This paper analyzes the achievable rates of a simple relay network by a semi-analytical approach considering ideal capacity achieving codes of appropriate rates and optimum symbol-by-symbol soft-output decoding at the relay. Applying the information combining technique, regions are illustrated where the upper bound provided by the max-flow min-cut theorem can be reached. Moreover, the results show that the lower bound obtained by the constraint of error-free decoding at the relay can be exceeded in many scenarios. Simulations with a nearly capacity achieving half-rate repeat accumulate code confirm the semi-analytic results.

Relay networks have attracted great attention in recent years due to their ability to extend the coverage of cellular systems. Their capacity is often analyzed by the max-flow min- cut theorem which provides an upper bound on the true capacity. However, the exact capacity is generally still unknown. This paper applies information combining to determine the achievable rate of a simple relay network. Presupposing an ideal code of appropriate rate and optimum symbol-by-symbol soft-output decoding at the relay, this semi-analytical approach delivers upper and lower bounds on the achievable mutual information of the network. The obtained results identify three different regions concerning the optimum rate of the code. Simulations of a nearly capacity achieving half-rate repeat accumulate code confirm the semi-analytic results.

Multicarrier transmission is an established technique for radio transmission systems and it can be considered as a promising approach for next generation wireless systems. In channel situations with high delay-spreads, OFDM requires large guard interval lengths which decrease the spectral efficiency significantly. Therefore, alternative multicarrier transmission systems are of great practical interest, which guarantee a reliable transmission without a guard interval. In multicarrier systems the subcarriers are weighted in a quite different way by the frequency selective channel, based on the inherent structure of the multicarrier transmission system. If the transmitter knows the channel, the subchannel's transmission parameters such as transmit power or modulation alphabets can be adapted. In this contribution, transmit power allocation schemes for BER performance improvements in a wavelet-based multicarrier transmission system with SVD equalization (singular value decomposition) are investigated under the constraints of a limited total transmit power and a given data rate. Thereby the differences between the optimal but highly complex solution and suboptimal but complexity-reduced solutions are analyzed. Numerical results are obtained by simulations assuming different time-variant channel characteristics, with delays larger than the multicarrier symbol duration.

In this paper, we propose a new algorithm separating blindly the sources of a flat MIMO-communication link, where all sources are independently channel encoded. To this end the proposed method exploits statistical dependencies caused by the channel code for blindly estimating the channel and simultaneously adjusting a linear equalizer. In contrast to most common blind source separation methods the suggested method completely resolves the phase and permutation ambiguity of the estimated channel impulse response and the equalizer. The performance of the presented method is evaluated by numerical results.

In recent publications the use of lattice-reduction for signal detection in multiple antenna systems has been proposed. In this paper, we adopt these lattice-reduction-aided schemes to the MMSE criterion. We show that an obvious way to do this is infeasible and propose an alternative method based on an extended system model, which in conjunction with simple successive interference cancellation nearly reaches the performance of maximum-likelihood detection. Furthermore, we demonstrate that, a sorted QR decomposition can significantly reduce the computational effort associated with lattice-reduction. Thus, the new algorithm clearly outperforms existing methods with comparable complexity.

Theoretical and experimental studies have shown that layered space-time architectures like the BLAST system can exploit the capacity advantage of multiple antenna systems in rich-scattering environments. We present a new efficient algorithm for detecting such architectures with respect to the MMSE criterion. This algorithm utilizes a sorted QR decomposition of the channel matrix and leads to a simple successive detection structure. The algorithm needs only a fraction of the computational effort compared to the standard V-BLAST algorithm and achieves the same bit error performance.

This paper presents two different code acquisition schemes for sectorized multi-antenna CDMA systems in downlink mode. The so-called "separated code optimization" (SCO) decomposes a two dimensional spreading code into its time and space component and treats each part separately. The SCO maximizes the signal to interference and noise ratio at each receiver. In contrast, the "joint code optimization" (JCO) does not distinguish between time and space component and acquires the code simultaneously. Both methods are evaluated by Monte Carlo simulations and their performance are compared.

In recent years, spectral efficiency has become a key parameter in mobile radio communications. For code division multiple access systems, high spectral efficiencies can only be achieved by the application of the multiuser detection schemes combating the inherent multiuser interference. This paper analyzes the systems performance of a quasi-synchronous OFDM-CDMA uplink for nonlinear successive interference cancellation (SIC). Realizing that channel estimation is crucial for the performance of interference cancellation schemes, their iterative structure is exploited for improving channel estimation as well as SIC significantly. First, it is shown that convolutional codes with low memory perform much better for extremely high system loads than codes with large memory. Furthermore, correlative, maximum likelihood (ML) channel estimation as well as iteratively improved ML estimation is compared. Simulation results demonstrate that the loss compared to the single-user bound with perfect channel knowledge amounts only 0.7 dB even for loads of /spl beta/ = 2, QPSK modulation and real channel estimation.

The use of multiple antenna systems provides a great performance advantage in Rayleigh fading environments. To exploit this advantage several schemes like the layered space-time architecture have been proposed, but mostly restricted to narrowband transmission. Recently, Lozano and Papadias have presented a generalization of the well-known V-BLAST detection algorithm for frequency selective fading channels. In order to improve the performance of this receiver, we propose an iterative detection algorithm in this paper.

We propose a "turbo" coding scheme for the multiple-input multiple-output (MIMO) Rayleigh fading channel consisting of the serial concatenation of a block code as the outer code and different orthogonal space-time block codes (STBC) for more than two transmit antennas as the inner code. Here, we consider the orthogonal STBC as a mapping scheme in space and time. At the receiver, we apply iterative space-time detection and decoding. We analyze the impact of different mapping strategies on the information transfer of the soft-input-soft-output (SISO) space-time detector. Moreover, we analytically show that additional performance gains over Gray mapping can be obtained by different mapping strategies. Furthermore, we use extrinsic information transfer characteristics (EXIT-charts) in order to predict the performance and the behavior of the system.

This paper addresses aspects of channel coding in orthogonal frequency-division multiplexing-code-division multiple access (OFDM-CDMA) uplink systems where each user occupies a bandwidth much larger than the information bit rate. This inherent bandwidth expansion allows the application of powerful low-rate codes under the constraint of low decoding costs. Three different coding strategies are considered: the combination of convolutional and repetition codes, the code-spread system consisting of one single very low-rate convolutional code and a serial concatenation of convolutional, Walsh-Hadamard and repetition code. The latter scheme is improved by combining the Walsh-Hadamard codes with an additional M-phase-shift keying modulation resulting in complex-valued Walsh-Hadamard codes (CWCs). Analytical performance evaluations will be given for these codes for the first time. The application of CWCs as inner codes in a serial code concatenation is also addressed. We derive a symbol-by-symbol maximum a posteriori decoding algorithm in the complex signal space in order to enable iterative decoding for the entire code. A comprehensive performance analysis by simulation of all the proposed coding schemes shows that the Walsh-Hadamard-based schemes are the best choice for low-to-medium system load. Note that even for fully loaded OFDM-CDMA systems, the concatenation with an inner complex-valued Walsh-Hadamard code leads to a bit-error rate less than 10/sup -5/ for an E/sub b//N/sub 0/ of about 6 dB.

The transmission quality of LEO (low Earth orbit) satellite communication systems suffers from shadowed links and the power limitation of the satellites. In order to combat these two impacts, the use of diversity techniques is normally a good solution. Because of the non-frequency selective behavior of the satellite channel, frequency diversity cannot be exploited. Furthermore, time diversity is limited due to the short delay times allowed for speech transmission. Under these circumstances the application of space diversity can improve the transmission quality enormously. A basic model of a LEO satellite concept is the Globalstar system. This system utilizes DS-CDMA technology (direct sequence-code division multiple access), in which space diversity is implemented by combining the convolutionally encoded signals from different satellites with the help of a RAKE receiver. The performance of this concept is compared to a new approach using turbo codes. Regarding the transmission via two satellites, the coded bits of the two coders are split between the two satellite channels. At the receiver iterative decoding can be implemented. This concept needs less processing time and it performs better than the Globalstar system in terms of the bit error rate.

The performance of linear multiuser detection (MUD) and nonlinear parallel interference cancellation (PIC) is analyzed for a quasi synchronous OFDM-CDMA uplink transmission. Specifically, we investigate two different approaches concerning the combination of linear MUD techniques such as the decorrelator and the MMSE approach with nonlinear parallel interference cancellation (PIC). It is pointed out that OFDM-CDMA systems offer great advantages over single carrier systems due to flat fading conditions on each subcarrier leading to much lower implementation costs of MUD techniques. Assuming perfectly known channel impulse responses for each user and a rough synchronization, it turns out that the application of the MMSE-MUD filter leads to an insufficient equalization of the channel. Therefore, we propose an approach where the PIC loop works directly on the channel output and the MMSE filter is only active in the initial iteration loop. With this approach, the combination of linear MMSE-MUD and PIC shows excellent performance even for very high system loads.

Wireless communication systems such as global system for mobile (GSM) communications are playing a growing role for data transmission. In order to ensure reliable transmission, the channel impulse response has to be estimated accurately. This is a difficult task particularly for fast fading channels caused by high-speed mobile units. We deal with the application of nonblind and blind channel estimation approaches to identify the full rate data Traffic CHannel (TCHF9.6) of GSM. We present a new iterative channel estimation scheme leading to a significant performance improvement especially for high-speed propagation with Doppler frequencies up to 500 Hz. Furthermore, it is shown that blind channel estimation schemes could be as efficient as nonblind methods when regarding bit error rates (BERs) after channel decoding in terms of the E~/sub b//N/sub 0/ ratio. Moreover, a solution for the scalar ambiguity inherent in all blind estimation approaches is suggested.

Channel estimation plays a leading role in wireless communication systems such as GSM (Global System for Mobile communications). Putting the main emphasis on investigating the application of non-blind and blind channel estimation approaches to the identification of a full rate data Traffic CHannel (TCH/F9.6), we present a new idea of an iterative channel estimation based upon the capabilities of channel coding. We show that this method leads to a significantly improved performance especially for high speed propagations with Doppler frequencies up to 500 Hz. Giving the bit error rates (BERs) before and after channel decoding in terms of the E~/sub b//N/sub 0/ ratio, we show, furthermore, that blind channel estimation schemes could be as efficient as non-blind methods.

Turbo codes have received great attention due to their outstanding performance. Unfortunately, a high performance is associated with large transmission delays, prohibiting an application for speech transmission. Hence, the aim of this paper is the comparison of turbo codes employing short interleavers with convolutional codes in terms of bit error rate performance and computational effort. Additionally, a pragmatic approach of bandwidth-efficient turbo-coded modulation is considered. Analyzing the structure of the transmitter and receiver, interesting results are presented concerning the design of the mapper. Furthermore, a new very simple soft-output demodulation algorithm is derived. In order to compare turbo codes with convolutional codes under realistic conditions, both are embedded in a direct sequence (DS) code division multiple access system. Besides this comparison, a compromise between a high coding gain (low code rate) and high direct-sequence spreading is worked out, including the consideration of the turbo-coded modulation scheme. Simulation results indicate that turbo codes with small block interleavers do not outperform conventional convolutional codes. Furthermore, it is shown that for coherent demodulation, low code rates and low DS spreading is superior to high code rates and high DS spreading.