Up to this day, mobile communications have evolved rapidly over the decades, mainly focusing on speed-up to meet the growing data demands of 2G to 5G. And with the start of the 5G era, efforts are being made to provide such various services to customers, as IoT, V2X, robots, artificial intelligence, augmented virtual reality, and smart cities, which are expected to change the environment of our lives and industries as a whole. In a bid to provide those services, on top of high speed data, reduced latency and reliability are critical for real-time services. Thus, 5G has paved the way for service delivery through maximum speed of 20Gbps, a delay of 1ms, and a connecting device of 106/㎢ In particular, in intelligent traffic control systems and services using various vehicle-based Vehicle to X (V2X), such as traffic control, in addition to high-speed data speed, reduction of delay and reliability for real-time services are very important. 5G communication uses high frequencies of 3.5Ghz and 28Ghz. These high-frequency waves can go with high-speed thanks to their straightness while their short wavelength and small diffraction angle limit their reach to distance and prevent them from penetrating walls, causing restrictions on their use indoors. Therefore, under existing networks it's difficult to overcome these constraints. The underlying centralized SDN also has a limited capability in offering delay-sensitive services because communication with many nodes creates overload in its processing. Basically, SDN, which means a structure that separates signals from the control plane from packets in the data plane, requires control of the delay-related tree structure available in the event of an emergency during autonomous driving. In these scenarios, the network architecture that handles in-vehicle information is a major variable of delay. Since SDNs in general centralized structures are difficult to meet the desired delay level, studies on the optimal size of SDNs for information processing should be conducted. Thus, SDNs need to be separated on a certain scale and construct a new type of network, which can efficiently respond to dynamically changing traffic and provide high-quality, flexible services. Moreover, the structure of these networks is closely related to ultra-low latency, high confidence, and hyper-connectivity and should be based on a new form of split SDN rather than an existing centralized SDN structure, even in the case of the worst condition. And in these SDN structural networks, where automobiles pass through small 5G cells very quickly, the information change cycle, round trip delay (RTD), and the data processing time of SDN are highly correlated with the delay. Of these, RDT is not a significant factor because it has sufficient speed and less than 1 ms of delay, but the information change cycle and data processing time of SDN are factors that greatly affect the delay. Especially, in an emergency of self-driving environment linked to an ITS(Intelligent Traffic System) that requires low latency and high reliability, information should be transmitted and processed very quickly. That is a case in point where delay plays a very sensitive role. In this paper, we study the SDN architecture in emergencies during autonomous driving and conduct analysis through simulation of the correlation with the cell layer in which the vehicle should request relevant information according to the information flow. For simulation: As the Data Rate of 5G is high enough, we can assume the information for neighbor vehicle support to the car without errors. Furthermore, we assumed 5G small cells within 50 ~ 250 m in cell radius, and the maximum speed of the vehicle was considered as a 30km ~ 200 km/hour in order to examine the network architecture to minimize the delay.
Cho Jae Ho;Cho Kwang Hwan;Keum Kichang;Han Yongyih;Kim Yong Bae;Chu Sung Sil;Suh Chang Ok
Radiation Oncology Journal
/
v.21
no.1
/
pp.82-93
/
2003
Purpose : To reduce the Irradiation dose to the lungs and heart in the case of chest wail irradiation using an oppositional electron beam, we used an Individualized custom bolus, which was precisely designed to compensate for the differences In chest wall thickness. The benefits were evaluated by comparing the normal tissue complication probablilties (NTCPS) and dose statistics both with and without boluses. Materials and Methods : Boluses were made, and their effects evaluated in ten patients treated using the reverse hockey-stick technique. The electron beam energy was determined so as to administer 80% of the irradiation prescription dose to the deepest lung-chest wall border, which was usually located at the internal mammary lymph node chain. An individualized custom bolus was prepared to compensate for a chest wall thinner than the prescription depth by meticulously measuring the chest wall thickness at 1 emf intervals on the planning CT Images. A second planning CT was obtained overlying the individuailzed custom bolus for each patient's chest wall. 3-D treatment planning was peformed using ADAC-Pinnacle$^{3}$ for all patients with and without bolus. NTCPS based on 'the Lyman-Kutcher' model were analyzed and the mean, maximum, minimum doses, V$_{50}$ and V$_{95}$ for 4he heari and lungs were computed. Results .The average NTCPS in the ipsliateral lung showed a statistically significant reduction (p<0.01), from 80.2${\pm}$3.43% to 47.7${\pm}$4.61%, with the use of the individualized custom boluses. The mean lung irradiation dose to the ipsilateral iung was also significantly reduced by about 430 cGy, Trom 2757 cGy to 2,327 cGy (p<0.01). The V$_{50}$ and V$_{95}$ in the ipsilateral lung markedly decreased from the averages of 54.5 and 17.4% to 45.3 and 11.0%, respectively. The V$_{50}$ and V$_{95}$ In the heart also decreased from the averages of 16.8 and 6.1% to 9.8% and 2.2%, respectively. The NTCP In the contralateral lung and the heart were 0%, even for the cases with no bolus because of the small effective mean radiation volume values of 4.4 and 7.1%, respectively Conclusion : The use of an Individualized custom bolus in the radiotherapy of postrnastectorny chest wall reduced the NTCP of the ipsilateral lung by about 24.5 to 40.5%, which can improve the complication free cure probability of breast cancer patients.
By the activation of ovary hormone, many morphological changes occur in the epithelial cell lines and muscle cells in rat uterus. These two cells in uterus are important to the implantation of embryo, maintaining pregnancy and starting parturition. One important change associated with the morphological change of these two cells in uterus is the change on prostaglandin(PG) metabolism. Its presence and synthesis in endometriurn and myometrium in uterus affects estrous cycle and the start of embryo implantation in uterus. It also performs as an important modulator in parturition. So the abnormally weak expression of PG causes difficulty during labor and over-expression causes pre-term labor. PG biosynthesis starts from either free or liberated arachidonic acids from membrane phospholipid by phospholipase. Such arachidonic acids are converted into PG catalyzed by Cyclooxygenase. Under normal physiological condition, Cyclooxygenase-1(COX-1) having 602 units of amino acids controls the synthesis of PG. It acts as a local hormone regulating vasomodulation of blood flow, flexible muscle movement, increasing the blood permeability and contributing the protective role in preserving integrity of the stomach lining and Cyclooxygenase-2 (COX-2) is induced by the inflammation, pregnancy and increased its expression until parturition. Lipid metabolite like PG is located in uterine and expression of COX-2 increased with pregnancy. Increased expression of COX proteins in epithelial cells and myometrial cells are told to increase the muscle contractility in uterus but decreased right after the labor in rat. It is a good sign indicating that COX proteins are deeply related to the start of labor. Currently, Several studies report the use of PG and COX-2 inhibitor as medication for controlled abortion or to prevent pre-term labor but they entail various side-effects. Our study proposed to suggest use of acupuncture as an another mediator to control abortion or pre-term labor without causing unnecessary side-effects by those medicines. Two acupuncture sites, LI-4 & SP-6 were selected due to their known efficacy. From the immunohistochemical staining of COX-2, normal expression of COX-2 protein in nonpregnant SD rat's uterus revealed that COX-2 protein was primarily detected in the lumina epithelial lining and in the epithelial cell lining contacting the stromal cells. High resolution optical microscopic scanning revealed distinguishable staining in the myometrial mucosa. LI-4 acupuncture administered nonpregnant rat's uterus showed strong expression for COX-2 in endometrium contacted with lumina epithelial lining of rat uterus and in myometrial mucosa. Stromal cells showed more staining than untreated nonpregnant rat's uterus and stronger staining in stromal cells contacting myometrial layer compared to untreated nonpregnant rat's uterus. SP-6 acupuncture administered nonpregnant rat's uterus showed weak expression for COX-2 in myometrial layers and stromal cells but no staining was visible in lumina epitheliai and glandular epithelial cells. Few stromal cells and myometrial mucosa were positively stained for COX-2. Pregnant SD rat's uterus was also immunostained for COX-2 expression after 18 days of pregnancy. Unlike to untreated nonpregnant rat's uterus, luminal epithelial cells were not positively stained for COX-2 but stronger staining for COX-2 was revealed in stromal cells. LI-4 acupunctured SD rat's uterus had very strong expression of COX-2 in luminal epithelial lining. Few stromal cells showed stronger positive COX-2 staining and myometrial layers also showed more expression than untreated pregnant rat. SP-6 acupuncture administered pregnant SD rat's uterus showed positive expression of COX-2 in epithelial cells of luminal mucosa layer but weaker than that of LI-4 acupuncture treatment's case. However, strong positive staining was revealed in stromal mucosa and myometrial layers. Virgin SD rat's uterus motility index during LI-4 acupuncture was 66.52 % (Prob〉T = 0.0197) compared to its motility before the acupuncture treatment but the motility index was slighdy elevated up to 79.58 % (Prob〉T = 0.1175) after the acupuncture. During the SP-6 acupuncture treatment for 30 minutes, uterus motility index was 90.52 % (Prob〉T = 0.1832) showing lesser decrement but consequently reached similar motility index decreasal to 79.95 % (Prob〉T = 0.0215) after the acupuncture treatment as LI-4 showed. LI-4 acupuncture tend to be a quick treatment to reducing the uterus motility in a virgin rat but eventually both two acupuncture administration created very similar reduction of uterus motility seeing the index after the both acupunctures. The uterus movement monitored during the LI-4 acupuncture administered for 30 minutes, Pregnant SD rat showed decreased motility down to 77.90 % (Prob〉 T = 0.0076) compared to uterus motility before the acupuncture and it continuously decreased down to 71.81 %(Prob〉T = 0.0214) after the removal of needle. The statistical analysis using paired t-test showed significance difference for both two motility indexs at =0.05. SP-6 acupuncture administered to pregnant SD rat also had similar pattern of decreasing uterus motility index down to 74.70 % (Prob〉T = 0.1730) during the initial 30 minutes acupuncture administration and it was continuously lowered to 71.52 % (Prob〉T = 0.0155) after the acupuncture. The paired t-test resuit for SP-6 suggest prompt response of uterus motility index to the SP-6 acupuncture treatment but consequently reached same level of inducing the motility reduction as LI-4 at =0.05 level.
Byambajav, Batkhuu;Alikhanov, Jumabek;Fang, Yang;Ko, Seunghyun;Jo, Geun Sik
Journal of Intelligence and Information Systems
/
v.24
no.1
/
pp.205-225
/
2018
Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.
Internet commerce has been growing at a rapid pace for the last decade. Many firms try to reach wider consumer markets by adding the Internet channel to the existing traditional channels. Despite the various benefits of the Internet channel, a significant number of firms failed in managing the new type of channel. Previous studies could not cleary explain these conflicting results associated with the Internet channel. One of the major reasons is most of the previous studies conducted analyses under a specific market condition and claimed that as the impact of Internet channel introduction. Therefore, their results are strongly influenced by the specific market settings. However, firms face various market conditions in the real worlddensity and disutility of using the Internet. The purpose of this study is to investigate the impact of various market environments on a firm's optimal channel strategy by employing a flexible game theory model. We capture various market conditions with consumer density and disutility of using the Internet.
shows the channel structures analyzed in this study. Before the Internet channel is introduced, a monopoly manufacturer sells its products through an independent physical store. From this structure, the manufacturer could introduce its own Internet channel (MI). The independent physical store could also introduce its own Internet channel and coordinate it with the existing physical store (RI). An independent Internet retailer such as Amazon could enter this market (II). In this case, two types of independent retailers compete with each other. In this model, consumers are uniformly distributed on the two dimensional space. Consumer heterogeneity is captured by a consumer's geographical location (ci) and his disutility of using the Internet channel (${\delta}_{N_i}$).
shows various market conditions captured by the two consumer heterogeneities.
(a) illustrates a market with symmetric consumer distributions. The model captures explicitly the asymmetric distributions of consumer disutility in a market as well. In a market like that is represented in
(c), the average consumer disutility of using an Internet store is relatively smaller than that of using a physical store. For example, this case represents the market in which 1) the product is suitable for Internet transactions (e.g., books) or 2) the level of E-Commerce readiness is high such as in Denmark or Finland. On the other hand, the average consumer disutility when using an Internet store is relatively greater than that of using a physical store in a market like (b). Countries like Ukraine and Bulgaria, or the market for "experience goods" such as shoes, could be examples of this market condition.
summarizes the various scenarios of consumer distributions analyzed in this study. The range for disutility of using the Internet (${\delta}_{N_i}$) is held constant, while the range of consumer distribution (${\chi}_i$) varies from -25 to 25, from -50 to 50, from -100 to 100, from -150 to 150, and from -200 to 200.
summarizes the analysis results. As the average travel cost in a market decreases while the average disutility of Internet use remains the same, average retail price, total quantity sold, physical store profit, monopoly manufacturer profit, and thus, total channel profit increase. On the other hand, the quantity sold through the Internet and the profit of the Internet store decrease with a decreasing average travel cost relative to the average disutility of Internet use. We find that a channel that has an advantage over the other kind of channel serves a larger portion of the market. In a market with a high average travel cost, in which the Internet store has a relative advantage over the physical store, for example, the Internet store becomes a mass-retailer serving a larger portion of the market. This result implies that the Internet becomes a more significant distribution channel in those markets characterized by greater geographical dispersion of buyers, or as consumers become more proficient in Internet usage. The results indicate that the degree of price discrimination also varies depending on the distribution of consumer disutility in a market. The manufacturer in a market in which the average travel cost is higher than the average disutility of using the Internet has a stronger incentive for price discrimination than the manufacturer in a market where the average travel cost is relatively lower. We also find that the manufacturer has a stronger incentive to maintain a high price level when the average travel cost in a market is relatively low. Additionally, the retail competition effect due to Internet channel introduction strengthens as average travel cost in a market decreases. This result indicates that a manufacturer's channel power relative to that of the independent physical retailer becomes stronger with a decreasing average travel cost. This implication is counter-intuitive, because it is widely believed that the negative impact of Internet channel introduction on a competing physical retailer is more significant in a market like Russia, where consumers are more geographically dispersed, than in a market like Hong Kong, that has a condensed geographic distribution of consumers.