Lukas Landau

Wireless interconnect for board and chip level

Electronic systems of the future require a very high bandwidth communications infrastructure within the system. This way the massive amount of compute power which will be available can be inter-connected to realize future powerful advanced electronic systems. Today, electronic inter-connects between 3D chip-stacks, as well as intra-connects within 3D chip-stacks are approaching data rates of 100 Gbit/s soon. Hence, the question to be answered is how to efficiently design the communications infrastructure which will be within electronic systems. Within this paper approaches and results for building this infrastructure for future electronics are addressed.

Robust adaptive beamforming algorithms using low-complexity mismatch estimation

We develop robust adaptive beamforming algorithms using low-complexity mismatch estimation (LOCME) of the array steering vector. The new feature is combined with the worst-case optimization-based beamformer, which can be solved as a second-order cone program (SOCP). In addition, we combine LOCME with a similar approach to the worst-case design using a joint optimization strategy based on the conjugate gradient method. The algorithms are developed in accordance with the constrained minimum variance design as well as the constrained constant modulus design criterion. While the LOCME feature does not affect the complexity order, the simulation results of all proposed algorithms show a superior performance, close to the optimum. Similarly to other mismatch estimation techniques, LOCME does not require additional information about the array steering vector mismatch. All proposed algorithms outperform the conventional worst-case optimization based algorithm, whereas the complexity is more than an order of magnitude lower, in case of the joint optimization-based approach.

A 60GHz LOS MIMO Backhaul Design Combining Spatial Multiplexing and Beamforming for a 100Gbps Throughput

In this work, a two-level hierarchical MIMO system is proposed to combine the spatial multiplexing gain and beamforming gain in a strong LOS channel. The superior is a MIMO system that consists of specially arranged sub-arrays to fully exploit the spatial multiplexing gain in deterministic channels. Additionally, a deterministic spherical-wave channel model is introduced. This channel model includes the radiation patterns of the sub-arrays, orthogonal phase relations introduced by the specific sub-array arrangement and the path loss considering deployment in practical scenarios. The attenuation includes the free space path loss, the oxygen absorption, the rain attenuation in bad weather and the front-end loss. The regulations for the maximum radiated power and the available bandwidth at 60 GHz were also investigated. Furthermore, the maximum transmission rate and upper bound of the energy efficiency are modeled and calculated for the proposed system operating at 60 GHz compliant to those regulations, as well for a constraint of the maximum available transmit power on-board. The result shows that the proposed system architecture is promising to achieve over 100 Gbps for macro-cell backhaul links with reasonable antenna sizes and high energy efficiency.

Information rates employing 1-bit quantization and oversampling at the receiver

The present paper deals with a bandlimited signal which is sampled with 1-bit precision at the receiver. The sampling rate is assumed to be an arbitrary integer multiple of the symbol rate. Assuming a scenario with additive Gaussian noise, pulse shaping filter and receive filter, the setup corresponds to a discrete channel with memory. The achievable rate is investigated by providing upper and lower bounds, where the latter correspond to different receiver principles. The numerical evaluation shows significant differences in the quality of the bounds. The results are valuable to evaluate system performance which confirm that oversampling with 1-bit precision is beneficial in terms of the achievable rate.

On reconstructable ask-sequences for receivers employing 1-Bit quantization and oversampling

Multigigabit communications based on 1-Bit quantization at the receiver benefit from relaxed power consumption as compared to conventional system designs, especially for short range communications. Utilizing oversampling at the receiver compensates partially the loss in terms of achievable rate brought by the coarse quantization via enabling and exploiting the inherent channel memory. This work proposes methods how to construct ASK-sequences which can be perfectly reconstructed at the receiver. Besides this advantage the proposed sequences are also superior in terms of achievable rate as compared to independent uniformly distributed input signals. Simulation based computations of the achievable rate confirm the theory of the model and the benefit of the oversampling approach.

Robust adaptive beamforming algorithms using the constrained constant modulus criterion

The authors present a robust adaptive beamforming algorithm based on the worst-case (WC) criterion and the constrained constant modulus (CCM) approach, which exploits the constant modulus property of the desired signal. Similar to the existing worst-case beamformer with the minimum variance design, the problem can be reformulated as a second-order cone programme and solved with interior point methods. An analysis of the optimisation problem is carried out and conditions are obtained for enforcing its convexity and for adjusting its parameters. Furthermore, low-complexity robust adaptive beamforming algorithms based on the modified conjugate gradient and an alternating optimisation strategy are proposed. The proposed low-complexity algorithms can compute the existing WC constrained minimum variance and the proposed WC-CCM designs with a quadratic cost in the number of parameters. Simulations show that the proposed WC-CCM algorithm performs better than existing robust beamforming algorithms. Moreover, the numerical results also show that the performances of the proposed low-complexity algorithms are equivalent or better than that of existing robust algorithms, whereas the complexity is more than an order of magnitude lower.

Energy-Efficient Transceivers for Ultra-Highspeed Computer Board-to-Board Communication

Enabling the vast computational and throughput requirements of future high performance computer systems and data centers requires innovative approaches. In this paper, we will focus on the communication between computer boards. One alternative to the bottleneck presented by copper wire based cable-bound communication is the deployment of wireless links between nodes consisting of processors and memory on different boards in a system. In this paper, we present an interdisciplinary approach that targets an integrated wireless transceiver for short-range ultra-high speed computer board-to-board communication. Based on our achieved results and current developments, we will also estimate energy consumption of such a transceiver.

Information Rates for Faster-Than-Nyquist Signaling with 1-Bit Quantization and Oversampling at the Receiver

Oversampling combined with low quantization resolutions has been shown to be a viable option when aiming for energy efficiency in multigigabit/s communications systems. This work considers the case of 1-bit quantization combined with oversampling and shows how the performance of such a system can be improved by using matched pulse shaping filters and faster than Nyquist signaling. The channel is considered with additive Gaussian noise and the performance of the system is evaluated in terms of achievable information rate under symbol-by-symbol detection.

1-Bit Quantization and Oversampling at the Receiver: Communication Over Bandlimited Channels With Noise

A bandlimited additive white Gaussian noise channel is considered where the output is 1-bit quantized and oversampled with respect to the Nyquist rate. We consider root raised cosine filters at the transmitter and receiver. In particular, we focus on a roll-off factor equal to 1 and 0. Because of the oversampling the channel has infinite memory. An auxiliary channel law is proposed which describes the resulting received sequences based on a truncated waveform. The random distortion due to the residual sidelobes can be considered as an additional noise term in the auxiliary channel law. The auxiliary channel law is utilized for computing a lower bound on the achievable rate and in a further step for optimizing a Markov source model. Different signaling schemes have been considered, such as BPSK and ASK. Moreover, Nyquist signaling and faster-than-Nyquist signaling are considered. The resulting achievable rates are superior as compared with results from the literature on bandlimited channels with noise, 1-bit quantization, and oversampling at the receiver.

Communication with 1-bit quantization and oversampling at the receiver: Spectral constrained waveform optimization

In case of wideband communications the analog-to-digital converter becomes a power consumption implementation bottleneck. Alternatively, transmission schemes based on coarse (1-bit) quantization and oversampling at the receiver can be beneficial. In this regard, information is conveyed in the zero-crossings. However, in presence of spectral constraints the waveform design becomes a challenge. In this work, faster-than-Nyquist BPSK signaling is considered, where runlength limited sequences are applied in order to engineer intersymbol interference. In addition, to further improve the achievable rate, the waveform is optimized by applying a suboptimal design criterion which corresponds to a convex optimization problem. A rate improvement of 10 to 20 percent by optimizing the waveform is observed, in comparison to Gaussian pulses, considering the spectral mask for the IEEE 802.11ad standard. For all cases, the simulation results show that using run-length limited input sequences is superior in terms of achievable rate as compared to independent uniformly distributed input symbols.