Within the hierarchy of the Software Defined Network (SDN) network stack, the control layer operates as the critical middleware facilitator of interactions between the data plane and the network applications, which govern flow routing decisions. In the OpenFlow implementation of the SDN model, the control layer, commonly referred to as a network operating system (NOS), has been realized by a range of competing implementations that offer various performance and functionality advantages: NOX [14], and ONIX [18]. In this paper we focus on the question of control layer resilience, when rapidly developed prototype network applications go awry, or third-party network applications incorporate unexpected vulnerabilities, fatal instabilities, or even malicious logic. We demonstrate how simple and common failures in a network application may lead to loss of the control layer, and in effect, loss of network control.To address these concerns we present the ROSEMARY controller, which implements a network application containment and resilience strategy based around the notion of spawning applications independently within a micro-NOS. ROSEMARY distinguishes itself by its blend of process containment, resource utilization monitoring, and an application permission structure, all designed to prevent common failures of network applications from halting operation of the SDN Stack. We present our design and implementation of ROSE-MARY, along with an extensive evaluation of its performance relative to several of the mostly well-known and widely used controllers. Rather than imposing significant performance costs, we find that with the integration of two optimization features, ROSE-MARY offers a competitive performance advantage over the majority of other controllers.

Lip cancers are relatively rare, but early diagnosis is important for a good outcome. Unfortunately, many patients experience a delay in diagnosis. A new machine learning method, deep convolutional neural networks (DCNNs), uses algorithms which can reportedly be used to classify dermatological diseases at the same standard as board-certified dermatologists. However, this has not been verified for locations such as the lips, scalp, and genitals.A DCNN was used to classify malignant (cancerous) and benign (non-cancerous) lip disorders and its performance was evaluated. The images in this study were taken from the photo database of Seoul National University Hospital (SNUH) in South Korea. To validate the results, additional images were collected from two other affiliated hospitals. A total of 1973 lip images from SNUH were used including 853 malignant and 1120 benign diseases.The DCNN was trained with 1629 images (743 malignant, 886 benign) and its performance was evaluated using testing and external validation sets containing 344 and 281 images, respectively. For comparison, 44 participants with different levels of training were asked to classify the images.The study found that the DCNN's performance was equivalent to the dermatologists, and was superior to the nondermatologists when classifying malignancy. When they referenced the DCNN result, non-dermatologists performed significantly better.Thus, DCNNs can be used to classify lip diseases at a standard equivalent to a board-certified dermatologist and they can help unskilled physicians to discriminate between benign and malignant lip diseases. DCNNs could therefore be used to improve diagnosis and consequent patient outcomes for those with suspected lip cancers. This is a summary of the study: Dermatologist-level classification of malignant lip diseases using a deep convolutional neural network This summary relates to https://doi.

In this paper, we propose a miniaturized implantable antenna exhibiting a broadside radiation pattern and wide operating bandwidth. Previously reported small implantable antennas often display omnidirectional radiation patterns which are not suitable for in-to-off wireless body area network. The proposed design overcomes this problem by optimizing the antenna structure inside a realistic brain implant environment, a seven-layer brain phantom including skin, fat, bone, dura, cerebrospinal fluid (CSF), gray and white matters. The seven-layer phantom was modeled in a full-wave simulation software, and then the antenna was embedded in dura layer. The antenna has a circular shape with a diameter of 10 mm and a thickness of 0.5 mm. The top and bottom insulating layers share the same dimensions of the antenna. With the given location and surrounding materials, the antenna geometry was optimized to resonate at 2.4 GHz and to radiate broadside. The optimal design was fabricated using a low-loss biocompatible PCB material, Taconic RF-35 (ε r = 3.5, tanδ = 0.0018), and tested in a seven-layer brain phantom implemented with semi-solid artificial tissue emulating (ATE) materials. The results of both the simulation and measurement revealed similar −10-dB impedance bandwidths of 13.8% and 14.9%, respectively, which are wider than those of most single-band implantable antennas operating at 2.4 GHz. The proposed antenna also displayed a measured peak realized gain of −20.75 dBi and an acceptable radiation efficiency of 0.24%, which are within the typical range. Furthermore, we calculated the specific absorption rate (SAR) and assessed its compliance with the IEEE safety guidelines.

A key feature of upcoming 4G wireless communication systems is multiple-input-multiple-output (MIMO) technology. To make the best use of MIMO, the antenna correlation between adjacent antennas should be low (< 0.5). In this context, we propose a correlation reduction technique suitable for closely spaced antennas (distance, d < λ/40). This technique reduces mutual coupling between antennas and concurrently uncorrelates antennas' radiation characteristics by inducing the negative group delay at the target frequency. The validity of the technique is demonstrated with a USB dongle MIMO antenna designed for LTE 700 MHz band. Measurement results show that the antenna correlation is reduced more than 40% using the proposed technique.

No abstract