Can We Exploit Machine Learning to Predict Congestion over mmWave 5G Channels?

It is well known that transport protocol performance is severely hindered by wireless channel impairments. We study the applicability of Machine Learning (ML) techniques to predict congestion status of 5G access networks, in particular mmWave links. We use realistic traces, using the 3GPP channel models, without being affected using legacy congestion-control solutions. We start by identifying the metrics that might be exploited from the transport layer to learn the congestion state: delay and inter-arrival time. We formally study their correlation with the perceived congestion, which we ascertain based on buffer length variation. Then, we conduct an extensive analysis of various unsupervised and supervised solutions, which are used as a benchmark. The results yield that unsupervised ML solutions can detect a large percentage of congestion situations and they could thus bring interesting possibilities when designing congestion-control solutions for next-generation transport protocols.

Dynamic Queue Synchronization for Congestion Control Over Wireless Sensor Networks

Congestion control over resource constrained sensor network is a critical challenge and it is also quite hard to fulfill the QoS constraints for end users. In this paper, a dynamic queue synchronization (DQS) scheme is introduced that can alter the data rate dynamically as per the current buffer length. Its performance is compared with existing schemes i.e., Dynamic Round Robin (DRR) and Random Early Discard (RED) using different routing protocols i.e., LEACH/PEGASIS/TEEN under various constraints (Throughput/End-to-End Delay/Routing Load/Energy).

A Probability-Based Clustering Algorithm with CH Election for Expanding WSN Life Span

Clustering achieves energy efficiency and scalable performance in wireless sensor networks (WSNs). A cluster is formed by several sensors nodes, and one of them is elected as Cluster-head (CH). A CH collects information from the cluster members and sends aggregated data to the base station or another CH. This article proposes a new clustering algorithm (EMESISC) that is based on each node's probability of becoming a CH. This node's probability depends on its residual energy, buffer length, and received signal power. We compared EMESISC with LEACH algorithm. Simulation results showed that EMESISC is superior to LEACH.

Intelligent Algorithm for Enhancing MPEG-DASH QoE in eMBMS

Multimedia streaming is the most demanding and bandwidth hungry application in today’s world of Internet. MPEG-DASH as a video technology standard is designed for delivering live or on-demand streams in Internet to deliver best quality content with the fewest dropouts and least possible buffering. Hybrid architecture of DASH and eMBMS has attracted a great attention from the telecommunication industry and multimedia services. It is deployed in response to the immense demand in multimedia traffic. However, handover and limited available resources of the system affected on dropping segments of the adaptive video streaming in eMBMS and it creates an adverse impact on Quality of Experience (QoE), which is creating trouble for service providers and network providers towards delivering the service. In this paper, we derive a case study in eMBMS to approach to provide test measures evaluating MPEG-DASH QoE, by defining the metrics are influenced on QoE in eMBMS such as bandwidth and packet loss then we observe the objective metrics like stalling (number, duration and place), buffer length and accumulative video time. Moreover, we build a smart algorithm to predict rate of segments are lost in multicast adaptive video streaming. The algorithm deploys an estimation decision regards how to recover the lost segments. According to the obtained results based on our proposal algorithm, rate of lost segments is highly decreased by comparing to the traditional approach of MPEG-DASH multicast and unicast for high number of users.

Accurate Model for Network-on-Chip Performance Evaluation Based on Timed Colored Petri Net

Network-on-Chip (NoC) is a power architecture that emerged to solve communication issues present in modern Systems-on-Chip (SoCs). NoC based architectures are very scalable and offer high levels of communication parallelism, among other features. Every efficient NoC implementation requires several design steps to accomplish indices of performance. Although there are many system level models, high-level models for NoC are representative in the context of design since they provide fast and accurate analysis, with low modeling effort, for further VHDL implementations. This work proposes a NoC model based on a Timed Colored Petri Net (TCPN) that computes performance indices seamlessly. Network latency and buffer occupation are of special interest in our approach as they represent the key indices when assessing NoC performance. As results, we have validated and refined the model of a 5×5 mesh NoC comparing its indices with equivalent VHDL RTL description under synthetic and real traffic situations. The proposed model is capable of analyzing the influence of the router service time on the average latency time, enabling internal NoC evaluation to optimize buffer length. Simulation results demonstrate the model suitability for latency evaluation with time estimation errors often below 1%. Furthermore, this paper discusses the effort required to extend the model with other NoC architectural features. We conclude that the use of a TCPN model of NoC generates accurate results providing as much detailed information as their equivalent experiments using VHDL description.

Posterior inferior cerebellar artery reimplantation: buffer lengths, perforator anatomy, and technical limitations

OBJECTIVE Reimplantation of the posterior inferior cerebellar artery (PICA) to the vertebral artery (VA) is a safe and effective bypass option after deliberate PICA sacrifice during the treatment of nonsaccular and dissecting aneurysms at this location. However, the anatomy and limitations of this technique have not been studied. The goal of this study was to define the surgical anatomy and buffer lengths specific to the proximal segment of the PICA related to 2 variations of PICA reimplantation: 1) reimplantation “along-VA” (simulating a dissecting VA aneurysm), and 2) reimplantation “across-VA” (simulating a nonclippable, proximal PICA aneurysm). METHODS Ten cadaver heads (20 sides) were prepared for surgical simulation. Twenty far-lateral approaches were performed. The PICA was mobilized and reimplanted onto the VA according to 2 different paradigms: 1) transposition along the axis of the VA (along-VA) to simulate a dissecting VA, and 2) transposition perpendicular to the axis of the VA (across-VA) to simulate a nonclippable, proximal PICA aneurysm. The buffer lengths provided by mobilization of the artery in each paradigm were measured and the anatomy of perforator branching on the proximal PICAs was analyzed. RESULTS The PICA was reimplanted in all surgical simulations. The most common perforating artery on the P1 and P2 segments was the short circumflex type. No direct perforator was found on the P1 segment. The mean buffer length with reimplantation along the VA axis was 13.43 ± 4.61 mm, and it was 6.97 ± 4.04 mm with reimplantation across the VA. The PICA was less maneuverable when it was reimplanted across the VA, due to perforator branches of the PICA (P3 segment). CONCLUSIONS The buffer lengths measured in this study describe the limitations of PICA reimplantation as a revascularization procedure for nonsaccular aneurysms in this location. PICA reimplantation is a revascularization option for dissecting VA aneurysms incorporating the PICA origin that are < 13 mm in length, and for nonsaccular proximal PICA aneurysms that are < 6 mm in diameter. The final decision to reimplant the PICA depends on careful inspection of perforator anatomy that is not visible preoperatively on angiography, as well as an assessment of technical difficulty intraoperatively.

A Consensus Algorithm with Buffer in Second-Order Multi-Agent with Time-Varying Delayed Communication

In this paper, we discuss consensus problems for a network of dynamic agents with fixed topologies. A consensus algorithm for multi-agent systems with time-varying delay is presented. With the consideration of the effects of network conditions, such as network-induced delays, packet dropouts, mis-sequence, etc., a communication buffer is introduced for avoiding agents’ communication error, and a method for choosing buffer length is proposed. Using this buffer design, received data is rearranged and transferred in the original order. With the presented consensus algorithm, agents’ consensus is well-performance, and all the agents match the average speed finally. Simulation oriented results verify the effectiveness of the proposed algorithm.