Summary With the increase in size of supercomputers, also increases the number of abnormal events. CPU overheating is one such event that decreases the system efficiency: when a CPU overheats, it reduces its frequency. This paper presents a machine learning solution to predict such events. The proposed algorithm is based on dynamic time warping for feature extraction and on a machine learning algorithm for classification. It predicts overheating events solely by analyzing the trends of the temperature of the CPUs and can deal with very low temperature sampling rates while having a negligible computational cost in practice. Our evaluation, using data coming from a production supercomputer, shows that the proposed solution can make predictions a few minutes in advance with a good accuracy. Furthermore, considering two simple preventive actions to avoid CPU overheating events, we present an analytical study that shows that our predictive solution is good enough to allow a significant reduction of the cost of overheating events.

Synchronization is a challenging problem especially for distributed systems, such as out-of-coverage D2D networks, as no common reference time is available. In such cases, devices use distributed synchronization algorithms, however, accurately determining when to stop the synchronization process is as challenging as achieving synchronization since they do not have the synchronization status of other devices in the network. From energy efficiency and performance perspective, the synchronization process should be stopped at all devices at the same time. In addition, to counteract the effect of propagation delays during synchronization, timing-advance (TA) clocks should be employed. This could be achieved in the synchronization process, however, after the devices are synchronized, there is no central mechanism to instruct them on TA clocks for transmitting or receiving data packets. In this paper, we propose a synchronization algorithm which, in an energy-efficient manner, allows devices to (i) acquire the synchronization status of others and terminate the synchronization process as soon as all devices in the network are synchronized, (ii) allow the synchronized devices to properly advance/regress their clocks prior to data communication by tracking their relative timing. We numerically demonstrate that the maximum synchronization error over multipath channels is around 0.6μs and it can be maintained during data communication.

Time synchronization is a key aspect of device-to-device (D2D) schemes, particularly in decentralized networks where no reference time is available. Distributed phase-locked loops (DPLL) is a synchronization algorithm well suited for decentralized situations. In this work, we study DPLL in the context of 5G networks, where we include in our analysis several practical aspects of D2D communication, such as propagation delays, multipath propagation, and the use of single-carrier frequency division multiple access (SC-FDMA). We propose practical methods to compensate for their effects, and introduce new performance metrics to evaluate the merits of the synchronization algorithm. Through simulations at the physical layer, which capture the effects of analog-digital conversions, we demonstrate that time synchronization in a decentralized setting is possible under the constraints specified by the 3GPP for D2D applications.

To ensure reliable communications in randomly varying and error-prone channels, wireless systems use adaptive modulation and coding (AMC) as well as hybrid ARQ (HARQ). In order to elucidate their compatibility and interaction, we compare the throughput provided by AMC, HARQ, and their combination (AMC-HARQ) under two operational conditions: in slow- and fast block-fading channels. Considering both incremental redundancy HARQ and repetition redundancy HARQ, we optimize the rate-decision regions for AMC/HARQ and compare them in terms of attainable throughput. Under a fairly general model of the channel variation and the decoding functions, we conclude that: 1) adding HARQ on top of AMC may be counterproductive in the high average signal-to-noise ratio regime for fast fading channels and 2) HARQ is useful for slow fading channels, but it provides moderate throughput gains. We provide explanations for these results which allow us to propose paths to improve AMC-HARQ systems.

LTE Device-to-device (D2D) discovery and communication has been considered a means to allow capacity offloading, range extension as well as commercial proximity services such as advertisement or social networking. Within this context, several service discovery approaches may be envisioned. In this paper, we study strategies which rely on a central function supporting service discovery between LTE UEs by push or pull mechanisms. We propose some simplified models to compare them in terms of control overhead and UE energy consumption. Then we propose and evaluate a new approach which combines both PUSH and PULL strategies and highlight the gain that can be achieved on service discovery performances.

To reduce the complexity of space-time multiuser detection, it has been proposed recently to exploit the spatial dimension by forming groups of users and apply the detection individually to each group. In this work we propose a new space-time receiver structure based on the group-optimal MMSE linear detector along with a new grouping algorithm that respects practical hardware limitations. Furthermore, an extension of the proposed structure which allows non-mutually exclusive grouping is presented. The simulation results show that the proposed reduced-complexity receiver structure provides a bit error rate (BER) performance close to the full linear MMSE multiuser detector.

Mobile communication systems based on DS-CDMA suffer from multiple access interference (MAI), which limits the system capacity. Several techniques, such as beamforming with multiple antennas and multiuser detection (MUD), have proved to be effective to mitigate MAI. To reduce the complexity of MUD, it has been proposed to process users in smaller groups. The interference between group is then reduced using beamforming with smart antennas. In this paper, a new group-based block linear successive interference cancellation structure is developed. The structure is shown to provide BER performance close to the full space-time MUD at a lower computational cost.

The increase in the demand for voice and data wireless services creates a need for a more efficient use of the available bandwidth. Current and future generations cellular systems based on DS-CDMA are known to be interference- limited. Several approaches exist to mitigate the multiple access interference including multiuser detection (MUD). Group-based techniques have been proposed to reduce the complexity of the MUD and have been shown to provide a performance-complexity tradeoff between match filtering and full MUD. In this work, we propose to reduce the inter-group interference (IGI), a limiting factor in group-based systems, using linear parallel interference cancellation (PIC). The complete equivalent matrix filter is derived and conditions for its convergence are discussed. The numerical results show that the proposed technique is effective against IGI, at a reduced computational cost.