學術出版;;Publications

Permanent URI for this communityhdl:11536/1

收集所有交通大學研究人員與學生個人或團體所發表之各項學術著作，涵蓋期刊論文、教師專書、學位論文等項。;;Collections of Publications of NCTU Researchers and Students Like Journal Papers and Thesis.

Browse

🏛️國立陽明交通大學機構典藏

Search Results

Now showing 1 - 10 of 53

Fast Packet Classification on OpenFlow Switches Using Multiple R*-Tree Based Bitmap Intersection
(2018-01-01)
Huang, Ding-Fong
;
Chen, Chien
;
Thanavel, Mahadevan
;
資訊工程學系
;
電機工程學系
In order to accomplish a stringent speed requirement for processing internet services such as Access Control List (ACL), Quality of Service (QoS), firewalls, etc., software based OpenFlow switches must have a fast packet classification capability. Even for hardware based OpenFlow switches, a limited size of Ternary Content Addressable Memory (TCAM) in the switch could be only enough for a forwarding table. Therefore, ACL, firewall tables, etc. need to be implemented by using the memory of the switch CPU. However, it has become a great challenge to build extremely effectively for next-generation software based packet classification that supports higher throughput and larger flow entries in OpenFlow switch. This paper first exploits a fast packet classification algorithm that forms a R*-Tree based Bitmap Intersection and secondly discusses an enhanced R*-Tree based Bitmap Intersection by using Bloom Filter and Multiple R*-Tree. The evaluation results show that the performance of the algorithm in OpenFlow switches is 4.42 times of Bitmap Intersection and 5.16 times of R*-Tree algorithm and consumes only 300 KB of memory space, which is much less than that of other methods. Finally, the use of multiple R*-Trees has further improved memory usage by about 30%.
Streaming Scalable Video Sequences with Media-Aware Network Elements Implemented in P4 Programming Language
(2018-01-01)
Wang, Guan-Ru
;
Chen, Chien
;
Chen, Chao-Wen
;
Pan, Li -Wen
;
Wang, Yu-Kong
We present the first Media-Aware Network Element (MANE) for intelligently streaming scalable video sequences in P4 programming language. Our MANE selectively drops queued scalable video packets when the queue occupancy exceeds a threshold. Three packet discarding logics are implemented: (i) tail, (ii) enhancement-layer, and (iii) rate-distortion optimized. Our P4-based MANE implementation is demonstrated in: (i) larger emulated networks in mininet with P4 software switches and (ii) a small real network with a physical P4 switch and multiple Raspberry Pis running P4 software switches.
A Zero Flow Entry Expiration Timeout P4 Switch
(2018-01-01)
He, Cheng-Hung
;
Chang, Brian Y.
;
Chakraborty, Suchandra
;
Chen, Chien
;
Wang, Li Chun
Existing OpenFlow based flow expiry mechanisms rely on a fixed timeout after which the switch proactively removes the flow entries from its flow table. Assigning an appropriate value to this timeout presents a tradeoff between efficient flow table utilization and controller overhead.We propose a simple addition of a zero flow entry expiration timeout scheme to idle timeout mechanism on a P4 based switch that identifies the last packet of a TCP flow, and removes the corresponding flow entry from the flow table. It enables connections to be deleted from the flow table right after it finishes, instead of waiting to expire after it reaches the timeout threshold and causing unnecessary flow table occupation.
Ictal networks of temporal lobe epilepsy: views from high-frequency oscillations in stereoelectroencephalography
(2019-10-01)
Peng, Syu-Jyun
;
Chou, Chien-Chen
;
Yu, Hsiang-Yu
;
Chen, Chien
;
Yen, Der-Jen
OBJECTIVE In this study, the authors investigated high-frequency oscillation (HFO) networks during seizures in order to determine how HFOs spread from the focal cerebral cortex and become synchronized across various areas of the brain. METHODS All data were obtained from stereoelectroencephalography (SEEG) signals in patients with drug-resistant temporal lobe epilepsy (TLE). The authors calculated intercontact cross-coefficients between all pairs of contacts to construct HFO networks in 20 seizures that occurred in 5 patients. They then calculated HFO network topology metrics (i.e., network density and component size) after normalizing seizure duration data by dividing each seizure into 10 intervals of equal length (labeled I1-I10). RESULTS From the perspective of the dynamic topologies of cortical and subcortical HFO networks, the authors observed a significant increase in network density during intervals I5-I10. A significant increase was also observed in overall energy during intervals I3-I8. The results of subnetwork analysis revealed that the number of components continuously decreased following the onset of seizures, and those results were statistically significant during intervals I3-I10. Furthermore, the majority of nodes were connected to a single dominant component during the propagation of seizures, and the percentage of nodes within the largest component grew significantly until seizure termination. CONCLUSIONS The consistent topological changes that the authors observed suggest that TLE is affected by common epileptogenic patterns. Indeed, the findings help to elucidate the epileptogenic network that characterizes TLE, which may be of interest to researchers and physicians working to improve treatment modalities for epilepsy, including resection, cortical stimulation, and neuromodulation treatments that are responsive to network topologies.
Early recovery of interhemispheric functional connectivity after corpus callosotomy
(2019-06-01)
Hung, Sheng-Che
;
Lee, Cheng-Chia
;
Chen, Hsin-Hung
;
Chen, Chien
;
Wu, Hsiu-Mei
Objective To investigate whether interhemispheric functional connectivity (FC) recovers in the first year after total callosotomy. Methods Eight epilepsy patients undergoing total callosotomy were recruited. Resting-state functional magnetic resonance imaging was acquired before and after surgery. The precallosotomy and postcallosotomy interhemispheric and intrahemispheric FC was analyzed by using graph theory and voxel-mirrored homotopic connectivity (VMHC). The seizure outcome was scored using the Engel surgical outcome scale. Results After callosotomy (mean postoperative interval = 4 months), the network density, average node degree, characteristic path length, and global efficiency of the whole interhemispheric networks were significantly decreased, compared to those in the precallosotomy networks. However, postcallosotomy interhemispheric FC and homotopic VMHC were not significantly reduced in bilateral frontal and temporal lobes. The network density and average node degree of the intrahemispheric networks were significantly increased. The characteristic path length and global efficiency of intrahemispheric networks were unchanged. Significance The interhemispheric FC may be preserved or recover early within the first postoperative year after total callosotomy, particularly in the frontal and anterior temporal lobes.
A Weighted ECMP Load Balancing Scheme for Data Centers Using P4 Switches
(2018-01-01)
Ye, Jin-Li
;
Chen, Chien
;
Chu, Yu Huang
;
資訊工程學系
;
資訊技術服務中心
This paper proposes a weighted Equal-cost multipath (W-ECMP) scheme in datacenters using a Programmable Data plane. Nowadays, data center networks face ongoing challenges for higher performance and greater efficiency. Equal-cost multi-path (ECMP) routing, the most common load-balancing scheme used on data center networks, tries to balance the traffic without awareness of the network congestion. To overcome this problem, CONGA and HULA are proposed successively. CONGA has implemented customized ASIC as a part of the switch fabric to collect congestion information distributively. Nevertheless, ASICs solution shows a lack of scalability and flexibility. However, although HULA solves some problems of CONGA by using p4 switch, best path selection strategy in HULA only selects and records the best path, which makes it easy to congest the best path, and HULA uses probes that waste bandwidth and deteriorate the performance. To solve the problems of HULA, W-ECMP chooses a path with a weighted probability to avoid congesting a path quickly. W-ECMP encapsulates the path congestion information into the normal traffic similar to the concept of Inband Network Telemetry (INT) in P4, which increases updating speed as the network loading increases. From experimental results, W-ECMP performs better than HULA does in average flow completion time (FCT).
SmartBone: An energy-efficient smart backbone construction in wireless sensor networks
(2007-07-01)
Chuang, Shun-Yu
;
Chen, Chien
;
資訊工程學系
;
Department of Computer Science
Wireless sensor network is a rapidly growing discipline with new technologies emerging, and new applications under development. The nodes in a wireless network generally communicate with each other along the same wireless channel. Unfortunately, sharing among wireless channels decreases network performance due to radio interference, and also raises energy consumption due to packet retransmission when interference occurs. Many topology control algorithms have been proposed to solve these problems. One widely used strategy is the backbone method. Backbone algorithms aim to reduce the backbone size. However, poor performance may be explored if only few backbone nodes are selected. Therefore, several heuristic algorithms such as SBC have been proposed. However, these algorithms cannot efficiently eliminate redundant nodes, and dramatically decrease performance, especially in relatively sparse networks. This study proposes a novel heuristic-based backbone algorithm called SmartBone to choose proper backbone nodes from a network. SmartBone includes two major mechanisms. Flow-Bottleneck preprocessing is adopted to find critical nodes, which act as backbone nodes to improve connectivity. Dynamic Density Cutback is adopted to reduce the number of redundant nodes depending on local area node density of network. SmartBone simultaneously considers the balance of network performance and energy savings. Significantly, the proposed algorithm has a 50% smaller backbone size than SBC, and improves the energy saving ratio from 25% using SBC to 40% using SmartBone. Moreover, SmartBone improves the packet delivery ratio from 40% to 90% when the density of sensor networks becomes relatively sparser.
Minimum-delay energy-efficient source to multisink routing in wireless sensor networks
(2007-12-01)
Chuang, Shun-Yu
;
Chen, Chien
;
Jiang, Chang-Jie
;
資訊工程學系
;
Department of Computer Science
This paper proposes a simple and scalable approach to multisink routing scheme in wireless sensor networks. Wireless sensor network is a rapidly growing discipline, with new technologies emerging and new applications under development. In addition to providing light and temperature measurements, wireless sensor nodes have applications such as security surveillance, environmental monitoring, and wildlife watching. One potential problem in a sensor network is how to transmit packets efficiently from single-source to multi-sinks, i.e., to gather data from a single sensor node and deliver it to multiple clients who are interested in the data. The difficulty of such a scenario is finding the minimum-cost multiple transmission paths. Many routing algorithms have been proposed to solve this problem. Most current algorithms address the reduction of power consumption, and potentially introduce a large delay. This paper proposes a novel multi-path routing algorithm, called hop count based routing (HCR) algorithm, which considers energy cost and transmission delay simultaneously. A hop count vector (HCV) is introduced to support routing decision. Moreover, an additional pruning vector (PV) can further enhance routing performance. The proposed algorithm also provides a maintenance mechanism to handle the consequence of faulty nodes. A failure of a node leads to an inaccurate HCV. Therefore, an efficient correction algorithm is necessary. An Aid-TREE (A-TREE) is applied to facilitate restricted flooding. This correction mechanism is more efficient than full-scale flooding for correcting the limited inaccurate HCVs. Finally, the impact of failed nodes is studied, and an algorithm, called Lazy-Grouping, is proposed to enhance the robustness of HCR. (C) 2007 Elsevier B.V. All rights reserved.
Network Aware VM Load Balancing in Cloud Data Centers Using SDN
(2017-01-01)
Tsygankov, Mykola
;
Chen, Chien
;
資訊工程學系
;
電機工程學系
;
Department of Computer Science
Despite being industry standard, load balancing via virtual machine (VM) migration methods in modern cloud data centers still exhibit two major performance issues: resulting system load balancing degree and total time till balanced state. Over the years research has been conducted to improve these metrics. Yet due to the fact that these issues influence each other, it has been a challenge to achieve substantial advancement in both metrics simultaneously. In this paper we study the problem of VM migration with load balancing and migration time constraints. We formulate this problem as a variation of multi-commodity minimum cost flow problem and show how both resulting degree of imbalance and a total migration time can be optimized together. We also propose a heuristic method based on Ant Colony Optimization family of algorithms. Through simulation results we show that our heuristic gives good performance results compared to optimal solutions to minimum cost flow problem. Furthermore, our method achieves on average twice faster migration time compared to existing methods, proposed in related works.
Dynamic Reordering Bloom Filter
(2017-01-01)
Chang, Da-Chung
;
Chen, Chien
;
Thanavel, Mahadevan
;
資訊工程學系
;
資訊技術服務中心
In order to check a membership in multiple sets of bloom filter in a dynamic bloom filter, a sequential search is usually used. Since the distribution of queried data is unpredictable because the distribution has a feature of temporal locality. Therefore more search cost is incurred if queried data is stored in the peer which is corresponded to the Bloom Filter has lower query priority. In this paper, we introduce Dynamic Reordering Bloom Filter that can save the cost of searching Bloom Filter by dynamically reorder the searching sequence of multiple bloom filters in a dynamic bloom filter with One Memory Access Bloom Filter (OMABF) and checked in the order saved in Query Index (QI). The performance of the system is evaluated by Markov Chain. Simulation results show that our scheme on average has 43% better in searching performance comparing with the sequential methods, which is verified via three different trace log files.

學術出版;;Publications

Browse

Filters

Settings

Sort By

Results per page

🏛️國立陽明交通大學機構典藏

Search Results