Christopher Knievel

Multi-Dimensional Graph-Based Soft Iterative Receiver for MIMO-OFDM

A graph-based receiver is presented that iteratively performs soft channel estimation and soft data detection. Reliability information of data symbols is utilized to improve channel estimation, and in turn, soft channel estimates refine data symbol estimates. The proposed multi-dimensional factor graph introduces transfer nodes that exploit correlation of adjacent channel coefficients in an arbitrary number of dimensions (e.g. time, frequency, and spatial domain). This establishes a simple and flexible receiver structure that facilitate soft channel estimation and data detection in multi-dimensional dispersive channels, and supports arbitrary modulation and channel coding schemes. Simulation results demonstrate that the proposed multi-dimensional graph-based receiver outperforms iterative and non-iterative state-of-the-art receivers.

On particle swarm optimization for MIMO channel estimation

Evolutionary algorithms, in particular particle swarm optimization (PSO), have recently received much attention. PSO has successfully been applied to a wide range of technical optimization problems, including channel estimation. However, most publications in the area of digital communications ignore improvements developed by the PSO community. In this paper, an overview of the original PSO is given as well as improvements that are generally applicable. An extension of PSO termed cooperative PSO (CPSO) is applied for MIMO channel estimation, providing faster convergence and, thus, lower overall complexity. Instead of determining the average iterations needed empirically, a method to calculate the maximum number of iterations is developed, which enables the evaluation of the complexity for a wide range of parameters. Knowledge of the required number of iterations is essential for a practical receiver design. A detailed discussion about the complexity of the PSO algorithm and a comparison to a conventional minimum mean squared error (MMSE) estimator are given. Furthermore, Monte Carlo simulations are provided to illustrate the MSE performance compared to an MMSE estimator.

2D Graph-Based Soft Channel Estimation for MIMO-OFDM

We address joint channel estimation and data detection based on factor graphs. The considered graph-based approach utilizes reliability information of channel estimates to facilitate soft-output data detection, and in-turn reliability information about the data symbols is taken into account for channel estimation. In this paper graph-based soft channel estimation and detection is extended to an OFDM based air interface, where the channel response varies in two dimensions; time and frequency. Initial channel estimates obtained by training symbols are conveyed by a two dimensional (2D) factor-graph in time and frequency with only a linear increase in complexity. The required training overhead for the proposed 2D graph-based soft channel estimation scheme may be substantially reduced by taking the redundancy introduced by the channel coding into account.

Particle Swarm Enhanced Graph-Based Channel Estimation for MIMO-OFDM

Iterative receiver structures that jointly perform channel estimation and decoding promise substantial performance gains. However, these gains only materialize with sufficiently accurate initial channel estimates. In this paper, initialization by multi-objective particle swarm optimization (MOPSO) is investigated. MOPSO supports low-complexity initial channel estimation with superimposed training symbols. Furthermore, it is shown that MOPSO works well in rank-deficient scenarios with arbitrary training sequences. Numerical results validate the performance enhancement of MOPSO initialization integrated within a graph-based iterative receiver.

Joint channel and parameter estimation for combined communication and navigation using particle swarm optimization

In this paper, a new method for joint channel and parameter estimation in the framework of combined communication and navigation is investigated. The basic idea is to estimate the parameters needed for positioning from the channel impulse response: In the estimator not only the channel coefficients of the equivalent discrete-time channel model are estimated, but also the parameters of the physical channel including the propagation delay of the line-of-sight path. A priori information about the pulse shaping filter and the receive filter are used. The estimator is based on the maximum-likelihood principle, which leads to a nonlinear minimization problem. The corresponding metric is minimized by particle swarm optimization, which is a simple global optimization algorithm that does not use any gradient information. The performance of the estimator is evaluated by means of Monte Carlo simulations. The results are compared to the Cramer-Rao lower bound and it is shown that the estimator is asymptotically optimal and efficient. The mean squared error of the channel estimates is decreased compared to the mean squared error of standard least squares channel estimates.

Improving Multi-Dimensional Graph-Based Soft Channel Estimation

The principles and benefits of soft decisions are well known and widely applied. The advantage of knowing that a decision is reliable or not is obvious. Belief propagation within a factor graph enables a unified use of soft information for both channel estimation and data detection. However, if soft information does not reflect the true reliability of a decision, the achievable performance may degrade. In this paper, the calculation of reliability information is refined to consider the event of unreliable soft decisions. The proposed solution is based on the mean bit error probability calculated after each iteration and integrates nicely within the existing factor graph. Simulation results are provided to illustrate the performance gains.

Evaluation and extension of a multi-dimensional graph-based receiver concept for MIMO-OFDM

In most modern wireless systems, adaptive modulation and channel coding is applied to achieve a high spectral efficiency. Hence, suitable receivers need to support a large variety of modulation and channel coding schemes (MCS) while maintaining a low complexity. The performance and complexity of a multi-dimensional factor-graph based (MD-GSIR) framework for iterative joint channel estimation and data detection is investigated for the application of adaptive modulation and coding in an LTE environment. Three detection algorithms within the MD-GSIR framework are investigated and a novel tree-based detection is proposed. The versatile structure of the MD-GSIR concept offers a flexible trade-off between computational complexity and performance. Furthermore, the message exchange is adapted to support the irregular training structure which may occur in adaptive environments.

Coded Sampling Bound—How Much Training is Needed for Iterative Semi-Blind Channel Estimation?

Coherent detection is commonly facilitated by means of pilot-aided channel estimation. While this approach offers a high estimation accuracy, the spectral and power efficiencies are reduced due to the pilot overhead. In the case that redundancy due to channel coding is not used for the purpose of channel estimation, the maximum pilot spacing for which a time-varying channel can be reconstructed without distortion is given by the Nyquist-Shannon sampling theorem. In this paper, it is shown that if channel coding is taken into account for channel estimation in an iterative fashion, the maximum spacing can be significantly extended, which results in an increased spectral efficiency. Towards this goal, a so-called coded sampling bound is derived. This semi-analytical bound is compared with the Nyquist-Shannon sampling theorem. The presented results indicate that the maximum pilot spacing can be arbitrarily large given a suitable channel code and code rate.

On the Combining of Correlated Random Measures with Application to Graph-Based Receivers

Nowadays, message combining is an essential component in most digital communication systems. Correlation between random measures has a significant impact on the combining process. In order to provide the best estimate after combining, correlation must be considered. In many applications correlation is obvious, e.g. correlation in the time, frequency, and/or spatial domain of a radio channel. In other cases, correlation is more concealed. In this paper, two methods to combine correlated random values are presented and applied to a graph-based iterative receiver. It is explained, why correlation in the message exchange arises and how it can be taken into account in the message combining step. Simulation results are provided showing the performance gains when correlation is considered.