Urbashi Mitra: Multihopped communication in wireless environments is receiving increased interest as a method by which to reduce transmission power and/or increase fidelity by reducing the attenuation of received signals. In this work, a new approach for multihopped, multicarrier signaling is examined. In particular, frequency-domain relay processing for two-hop transmission is investigated. Due to the use of orthogonal frequency division multiplexing (OFDM), amplify-and-forward relay processing can be generalized to allow for sub-carrier permutations of the information streams at the relay. By further optimizing the transmission power, this multicarrier amplify-and-forward offers strong performance with moderate complexity in contrast to the extremes of decode-and-forward and conventional amplify-and-forward. Although the end-to-end achievable rate for this scenario is a non-concave function of the power allocation vectors, we can determine an algorithm whose computational complexity grows quadratically with the number of sub-carriers. Furthermore, we develop a linear complexity algorithm that achieves near optimal performance. For hops with a frequency-flat channel response, the maximum achievable rate is explicitly derived from the associated optimization. For hops with Rayleigh fading frequency-domain channel responses, numerical results are presented and it is illustrated that the proposed low-complexity algorithm usually achieves near-optimal performance.
Joint work with Wenyi Zhang and Mung Chiang
Joint work with Rudolf Ahlswede, Igor Bjelakovi\'c, Janis Nötzel, Rafel Wyrembelski
Wolfgang Utschick: We consider MIMO interference networks, in which each user transmits a single beam, and examine the performance at high SNR as characterized by the asymptote of the achievable rate over the logarithmic SNR. This asymptote can be described by its slope (corresponding to the degrees of freedom of the system) and offset. It is known that interference alignment achieves the optimal slope; the alignment conditions, however, fully constrain the system and leave little room for optimizing the offset. Transmitting without interference alignment, i.e., enabling only N users, where N is the number of antennas at each receiver and transmitter, achieves only about half the optimal slope, but a potentially far higher offset, meaning that the asymptotes of the aligned and the non-aligned strategies cross at some point. We attempt to quantify this trade-off between asymptotic slope and offset in a Gaussian i.i.d. channel model by means of some large system approximations and discuss the behavior for increasing system dimensions.
Joint work with David Schmidt and Michael L. Honig
Eduard Jorswieck: The characterization of the Pareto boundary of the multi-antenna interference channel is necessary in order to compute efficient operating points. Motivated by the two-user beamforming in multi-antenna interference channels, we characterize the upper boundary of the so-called achievable single-user gain-region. The eigenvector corresponding to the maximum eigenvalue of the weighted sum of Hermitian forms of channel vectors is shown to achieve all points on the in some given direction. We show that these eigenvector achieve also all points of the boundary of the K-user interference channel achievable rate region. This result can be applied to a large class of interference network scenarios, including the MISO interference channel, the multicast beamforming, and the beamforming for private messages. Numerical simulations illustrate the achievable gain-region.
Slawomir Stanczak: An admission control algorithm for power-controlled wireless networks, proposed previously for the case of linear interference functions, is considered in this paper. We analyze the properties of the algorithm using the framework of standard interference functions, which makes it applicable to many system designs. Furthermore, we introduce individual power constraints into the system. The key property of the algorithm is the protection of active users, which guarantees that as new users attempt to join the network, the quality of the established links is sustained. We present conditions under which this key property is preserved under power constraints and analyze the convergence properties of the scheme.
Maxime Guillaud: The object of this contribution is to propose a comparison between various linear precoding schemes in the context of the MIMO flat-fading Gaussian interference channel. Performance (in terms of the achieved mutual information) of codebooks based on joint linear precoders design (such as interference alignment) is compared to the performance achieved through distributed (based on game-theoretic principles) interference management techniques, based on analytical expressions whenever available, and on Monte-Carlo simulations.
Joint work with Roland Tresch
Deniz Gündüz: We study the problem of simultaneous multicasting of multiple messages with the help of a relay terminal. In this model, termed as the compound multiple access channel with a relay (cMACr), a relay station simultaneously assists two transmitters in multicasting their independent messages to two separate receivers. The relay may also have an independent message of its own to multicast. We first provide achievable rate regions based on decode-and-forward (DF) and compress-and-forward (CF) strategies. Then, we derive an outer bound for the special case, called the cMACr without cross-reception, in which each transmitter has a direct link to one of the receivers while the connection to the other receiver is enabled only through the relay terminal. The capacity region is characterized for a binary modulo additive cMACr without cross-reception, proving the optimality of binary linear block codes in this setup, and hence, highlighting the benefits of physical layer network coding and structured codes. We then extend our results to the Gaussian channel model, providing achievable rate regions for DF and CF, as well as for a structured code design based on lattice codes. We show that the performance with lattice codes approaches the upper bound for increasing power, surpassing the rates achieved by the considered random coding-based techniques.
Joint work with Osvaldo Simeone, Andrea J. Goldsmith, H. Vincent Poor and Shlomo Shamai (Shitz)
Zoran Utkovski: Motivated from results for the point-to-point block MIMO channels without channel knowledge, communication schemes for the non-coherent two-way relaying channel (no channel knowledge at the relays and the terminals) are investigated. A construction from packings in Grassmann manifolds is proposed and discussed.
Aydin Sezgin: The achievable degrees of freedom (DOF) in an interference relay channel are examined. We highlight some differences between two relaying strategies in two different setups. Namely, we study the DOF achieved using interference alignment and interference nulling in a cognitive relay setup and in a half-duplex relay setup. The difference between the two setups is causality, which leads to a difference in the achievable DOF. We show that 3/2 DOF are achievable almost surely in a 3-user complex Gaussian interference cognitive-relay channel with quasi-static channel coefficients and with one antenna at each node. On the other hand, we show that with a cognitive relay, interference nulling outperforms interference alignment, since it achieves 2 DOF almost surely. For the half-duplex setup, it was shown that 3/2 DOF of freedom are achievable if the channel is time-varying by considering the 2 symbol extension equivalent MIMO setup. The achievability of 3/2 DOF in the equivalent MIMO is compromised if the channels are quasi-static. In this case, 6/5 DOF are achievable using interference alignment with asymmetric complex signaling, while interference nulling can not achieve 1 DOF almost surely.
Joint work with Anas Chaaban.
Gerhard Wunder: In this talk we introduce a class of autonomous network controllers for wireless networks with multiple antennas. Based upon utility maximization framework distributed network controllers collect all local available information and a small amount of global message exchange and steer the network into a new operating point governed by the underlying utility measure and a suitable virtual model. We will discuss several issues regarding this approach such as stability, quantized feedback etc. and evaluate it by simulations.
Joint work with A. Stolyar, H. Viswanathan, and M. Kasparik
Victor Zyablov: There are N stations, which can transmit data. The data of each station is divided into success blocks of length n and each station is in advance provided with its transmission protocol. We denote the protocol of the ith station by gi=(gi1,K,gin). If the position of gi contains 0 (gij=0), this means that ith station is silent in this position; if contains 1, this means that ith station transmits one of the binary symbols a or b. Access to the channel is restricted; simultaneously, at most S stations out of N have access to the channel during time T (T ≥ n). It is assumed that there is no block synchronization between stations (there is only symbol synchronization). At the channel output, we obtain a sequence y = (y1, y2,... , yT) of length T over the quaternary alphabet: 0, a, b, and x. We consider the following model of output sequence generation: y_t= 0 if at time t all s stations were silent;
a or b if at time t on station was transmitting a symbol,
and this symbol was a or b, respectively;
x if at time t two or more stations out of s were transmitting