This paper discusses the application of long short-term memory (LSTM) neural networks to maintain the synchronization of a real-time clock in holdover operation, that is, while the timing reference input of the clock is unavailable. The approach trains the LSTM network based on timestamps acquired while the slave clock is locked to its reference input coming from a master clock. When the slave clock loses its reference and enters holdover mode, the LSTM takes over and controls the clock. We evaluate the method on a testbed consisting of IEEE 1588 Precision Time Protocol (PTP) clocks based on field-programmable gate arrays (FPGA), where we collect nanosecond-accurate timestamps for offline analysis. We evaluate two oscillator stability scenarios: when the PTP clocks rely on oven-controlled crystal oscillators (OCXOs) and when they use crystal oscillators (XOs). In both cases, we demonstrate that the algorithm can sustain the clock synchronization accuracy within reasonable limits over intervals of 1000 seconds in two different temperature scenarios.

The 5G mobile communication systems introduce multiples functional splits between base station elements, new transmission bands and a large number of antennas supporting beamforming. In this scenario, a viable strategy to avoid building penetration losses is to deploy the antenna elements indoor and use a fronthaul link to establish the connection between them and the rest of the radio access network. This work explores a 5G fronthaul scheme based on analog radio over coaxial cables, leveraging this existing fixed access infrastructure to facilitate 5G deployments. Results indicate that the fronthaul solution discussed here is able to support the high capacity requirement of 5G, with the additional benefits of low transmit power, low latency and low cost. The main novelty of this paper is to investigate the potential of this kind of fronthaul architecture considering both the physical layer and economic aspects.

Massive Beamforniing systems in next-generation Centralized Radio Access Network architectures can substantially increase the requirements on fronthaul interfaces. This paper proposes a novel and efficient model to reduce the overhead generated by the transfer of beamforming coefficients between base station elements. This approach creates a compressed version for said coefficients, in the beamspace domain, using a clustering strategy over sets (windows) of most powerful beams. The method can be extended for any antenna configuration and also parameterized to adjust the trade-off between radio performance and the fronthaul load reduction. Simulations in typical mobile outdoor scenarios show that the proposed method can present Signal-to-Interference-plus-Noise-Ratio performance comparable to the usual dimensionality reduction approach, i.e., beam selection, while generating lower fronthaul overhead.

Crosstalk cancellation is a crucial issue for traditional digital subscriber line systems. For LTE-over-copper systems, however, the need for crosstalk cancellation varies depending on the analog fronthauling architecture and its parameters. A crosstalk handler that is decoupled as much as possible from the rest of the system architecture is thus preferred. Therefore, we propose a time-domain precoding scheme specifically in downstream direction to separate the precoding unit from the LTE signal flow, and perform crosstalk cancellation in an on-demand manner. Estimation of all direct and crosstalk paths is assisted by LTE cell-specific reference signals. The time-domain precoder transforms the interference channel into a crosstalk-free channel with identical direct paths, which allows for low-complexity inter-symbol interference mitigation. We evaluate the concept in terms of signal-to-noise ratio provided for LTE signals using measured wireline channel data.

Next-generation wireline systems may exploit fre- quencies up to several hundred MHz on short lines. Strong crosstalk coupling, comparable to the direct paths, is one of the main channel characteristics at high frequencies. Instead of fully cancelling all crosstalk, we utilize strong crosstalk paths to boost data-rate for active users. Two linear precoding schemes, based on maximum ratio-combining (MRC) and convex optimization, respectively, are proposed and applied to a common network topology. The precoding schemes exploit constructive crosstalk signals on unused lines or in unused parts of the spectrum on neighboring lines to boost data-rate while still complying with the regulated spectral power limits per line. More than 500 Mbps throughput gain per active user can be achieved compared to state-of-the-art linear precoding.

To increase indoor coverage for mobile services, we propose a residential small cell infrastructure making use of the existing copper plant. The system is cabinet-based, collocated with VDSL2 and uses small pieces of spectrum next to VDSL2. Inspired by the Ericsson Radio Dot System, it challenges the femtocell paradigm offering full macro functionality in the small cells. An interesting service potential is offered albeit the added mobile traffic capacity is moderate as it is limited by the copper fronthaul.

Vectoring is a vital component of wideband wireline communication systems. A disorderly leaving event (DLE) disturbs the vectoring operation since the precoder, which was designed for the channel before the change, is no longer up to date. Measurements indicate that the impact of a DLE can be serious for frequencies beyond 30 MHz, which corresponds to the band used by emerging wideband communication systems over short multi-pair copper cables such as G.fast. As an alternative to the state-of-the-art update procedure, this paper presents an approach to mitigating the DLE problem. By interpreting DLE with the FEXT-reflected-NEXT (FRN) model, we propose a scheme that enables the showtime lines to return to disturbance-free transmission once the loss of signal on a certain line is detected while updating the precoder as a background process. Furthermore, the estimation complexity for a K-user vectoring group is reduced from O(K2) to O(K).

The densification of mobile networks in order to meet increased capacity demands is ongoing, needed and costly. A few papers have been published based on the insight that the fixed broadband networks offer a multitude of sites, for instance our homes, for potential small cell deployment providing backhaul capacity and power without site costs. However, in order to reach economical large-scale benefits, we explore the case when radio systems are deployed in coexistence with DSL. In this paper, we establish the feasibility of such a concept under constraints invoked by state-of-the-art and emerging systems (3GPP, VDSL2 and G.fast) and make statements about the required architecture. We also point out that the enthusiasm of previously published results should be lowered a notch.

Emerging wireline transmission systems such as G.fast use bands up to around 200 MHz on short cables. A key enabler for achieving the aspired throughput of several hundred Mbit/s is joint processing of transmit signals in downstream direction as well as joint processing of receive signals in upstream direction through techniques referred to as vectoring. A new challenge in such systems are sudden and severe changes in the channel matrix caused by changing terminations on lines outside the vectoring group. Such events can be caused by users disconnecting their modems, turning them on or off, or on-/off-hook events on lines that still support the plain old telephony service. This work presents channel measurements capturing the impact of termination changes caused by modems or handsets. An analysis of the impact of these sudden changes on the signal-to-noise-power-ratio in vectoring systems reveals that throughput and stability can be seriously degraded. The potential of decision-directed channel tracking based on least squares estimation is investigated.

Cooperation of transceivers on signal level at the remote terminal-often referred to as vectoring-is an effective approach to mitigate crosstalk in wireline transmission systems. While the gain in bitrate compared to crosstalk-limited transmission is substantial, the initialization of a vectoring system requires time and care to ensure system stability. In this contribution, a fast initialization scheme for vectored wireline systems based on spectral protection is analyzed. Two scenarios are considered, where in the first one no prior channel information is available, and in the second one erroneous prior knowledge of channel coefficients is utilized. Measurements on a commercial VDSL2 vectoring platform and throughput simulations demonstrate the potential of the fast initialization scheme.