Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science 92 (2016) 493 – 497
2nd International Conference on Intelligent Computing, Communication & Convergence (ICCC-2016) Srikanta Patnaik, Editor in Chief Conference Organized by Interscience Institute of Management and Technology Bhubaneswar, Odisha, India
Portable Network Monitor using ARM Processor A. Monasa*, A. Vermaa, A. Gawaria, Mrs R. S. Paswanb b
a Student, PICT College Pune 411043, Asst. Professor, PICT College Pune 411043
Abstract The ever increasing demands for energy and cost reduction of networking devices and servers is driving the scientific and industrial communities to take in deeper considerations over the hardware and software techniques deployed in making these accessories. The best approach is to have parallel simultaneous development of the software and hardware as well rather opting for development of any one of the aspect from software or hardware of a system. For having reduced cost but same hardware effect over a system, the upcoming smart processors like ARM can handle well amount of loads and the software part needs to be fabricated in the manner to support such System-On-Board (SOB) mechanisms. This paper defines a scope for developing a network monitor which is established over traditional computers on just a credit card sized small computer. Basic idea is to replace general systems with dedicated systems for achieving power consumption and cost reduction goals. © byby Elsevier B.V.B.V. This is an open access article under the CC BY-NC-ND license © 2016 2014The TheAuthors. Authors.Published Published Elsevier (http://creativecommons.org/licenses/by-nc-nd/4.0/). Selection and peer-review under responsibility of scientific committee of Missouri University of Science and Technology. Peer-review under responsibility of the Organizing Committee of ICCC 2016 Keywords: ARM processor, Embedded web server, Traffic classification, Packet clustering, Internet protocol, Android platform.
* Corresponding author. Mr. Abhishek Monas, Tel.: +91-9673625259 E-mail address:
[email protected]
1877-0509 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the Organizing Committee of ICCC 2016 doi:10.1016/j.procs.2016.07.373
494
A. Monas et al. / Procedia Computer Science 92 (2016) 493 – 497
1. INTRODUCTION To develop a network Monitor model over ARM processor as a handy tool for specific network administrators, new technologies are proving the secure significance of networking domain and embedded systems[1]. Now we have much smarter network tools for administration and prolonged security methods by which we can have even firm grip over the essential uses and exploitation of the network resources. But these tools are either software or hardware belongings of admin domain are also competing to acquire less space and voracity by the physical means. So, today much more dedicated systems are developed to support specific and integrated tasks pertaining to dedicated loads like if you want to have a computer just for reading books then better go with kindle. Things to note here is to only develop applications supporting concurrency or sequential form of execution as BBB offers only one core and it will not ever give response to parallel programming. This all comes to its compiler’s last stage where all the parallel executions are converted to sequential executions [6].
Figure 1 System Flow
2. SETTING UP THE RIGHT BOARD Family of SOBs Different families like Arduino, Beagle Bone offers a great range of different SOBs but the most preferred one is Beagle Bone Black (BBB) with ARM processor A8 cortex as it is successor of Beagle Bone
A. Monas et al. / Procedia Computer Science 92 (2016) 493 – 497
White and leads in the count to handle memory and processor capacity over other families [1]. Other reason to go with BBB are as follows [2], • Remote communication interfaces • Multiple communication protocol support • High Level Operating System support • Real-Time capabilities Setting up a network monitor on BBB BBB can behave as a smart node of any network which has its range of applications from HPC through clustering till small network node as reference. In cluster model, the cluster is composed of five BeagleBoard-xM connected through a 100 Mbit Ethernet interface [2]. The CPU speed alteration provides the best resource utilization on BBB. One board acts as head node and serves as a gateway for managing the cluster and monitoring its status. The remaining boards are worker nodes entirely dedicated to data processing [4]. In a reference model BBB can simply act as a network edge and participate in different functionalities [5]. In network monitor model, we are using BBB as a node element whereby BBB can collect all the packets and create a pcapng file, unique for every network and time interface using tools like tcpdump, wireshark or tshark if you are only using command prompt or terminal [7]. These pcapng files can be recursively parsed and every packet details can be saved to readable .txt or .csv files by the available python libraries like dpkt. Now these packets or network frames can be used for network traffic classification using database and simple on board servers. This flow is presented in Fig 1. The information processed after parsing the pcapng files and stored into .csv file is shown in Fig 2. 3. CLASSIFICATION AND DISPLAY (FRONTEND) Earlier network traffic classification using traditional techniques such as port number based and payload analysis based techniques are no more effective because various applications and third party software uses port hopping and encryption technique to avoid detection like torrent client and skype [7]. Recently machine learning techniques such as supervised, unsupervised and semi supervised techniques are used to overcome the problems of traditional techniques. We prefer to use semi supervised machine learning approach to classify the network traffic using DBSCAN algorithm. This techniques uses only flow statistics to classify the network traffic [3]. This methodology is based on machine learning principle, consists of two components: clustering and classification. The goal of clustering is to partitions the training data set in to two disjoint group flow and traffic class. After making clusters classification is performed in which labeled data are used for assigning class label to the cluster [3]. After classification, different network parameters are derived like frame time, relative frame data protocol analysis, associated MAC addresses to the allocated IP addresses, total bandwidth consumption by each node, different protocols used in the communication, etc. We call all this data as final statistical results. This data is stored in csv (comma separated values) files for two reasons. First: it provides easy compatibility of data with various tools and second: csv files have got standard databases support like Mysql, MongoDB etc. The efficiency factor of any ARM processor is solely dependent on the type of instructions used over its hardware. Concurrent programming is best suitable for ARM architectures as it bares only one core so no scope for parallel executions. ––
495
496
A. Monas et al. / Procedia Computer Science 92 (2016) 493 – 497
Figure 2. Output of Parser
This statistical results can be stored over the embedded servers creating multiple database interfaces [2]. Server can be created on BBB itself. From this server all the device will access data link and statistical results. Server will be robust and has strong security constraints, it can only be accessed through key provided to the operator. Server can be created with professional tools like Ngrok. Ngrok secures introspectable tunnels to localhost web hook development tool and debugging tool. Ngrok are used here for tunneling of localhost of embedded system to a particular server. For the front end display unit, anything can be used to display the results, i.e. Virtual Network Connection (VNC) can be made from any display unit to show the results. Different open source models are preinstalled in some open source OS like Fedora (14 or higher versions). In our model we have used web view feature from Android Development Tools(ADT) and made a small application for this purpose. Android Application calls the stored links and accesses the data stored on BBB server through Ngrok. Please refer Fig 3 for a better understanding of the Frontend User Interface.
Figure 3. Display (Frontend)
A. Monas et al. / Procedia Computer Science 92 (2016) 493 – 497
4. CONCLUSION By the proposed paper and the related work in the embedded and networking domain, an attempt is made to redesign the computational requirements through dedicated architectures so as to make the overall system cheaper with increased efficiency and thus making sure to utilize the maximum possible resources like power, computation speed, memory available with the systems. To achieve best efficiency concurrent programming is suggested. Here using BBB as a ready board with mounted ARM processor, CPU speed is shifted from high to low according to the associated work, i.e. Classification and clustering needs much power rather than just displaying the results. Hence making sure to obtain the best power consumption and efficiency. Using small dedicated systems for a particular use also ensures the cost reduction aspects. References 1. 2. 3. 4. 5. 6. 7. 8.
Cleva, S., L. Pivetta, and P. Sigalotti. "Beaglebone for embedded control system applications." proc. ICALEPCS2013, San Francisco, CA, USA(2013). Neagoe, Teodor, Ernest Karjala, and Logica Banica. "Why ARM processors are the best choice for embedded low-power applications?." 2010 IEEE 16th International Symposium for Design and Technology in Electronic Packaging (SIITME). 2010. Shaikh, Mr Shezad, Mr Niket Bhargava, and Ms Urmila Mahor. "Implementation Of Internet Traffic Classifier Using Dbscan Algorithm."algorithms 2.5 (2012). Principi, Emanuele, et al. "Low power high-performance computing on the Beagleboard platform." Education and Research Conference (EDERC), 2012 5th European DSP. IEEE, 2012. He, Nannan, Han-Way Huang, and Brian David Woltman. "The Use of BeagleBone Black Board in Engineering Design and Development." (2014). Leupers, Rainer. "Compiler design issues for embedded processors." Design & Test of Computers, IEEE 19.4 (2002): 51-58. Nguyen, Thuy TT, and Grenville Armitage. "A survey of techniques for internet traffic classification using machine learning." Communications Surveys & Tutorials, IEEE 10.4 (2008): 56-76. Bernard, T., et al. "A general model of concurrency and its implementation as many-core dynamic RISC processors." Embedded Computer Systems: Architectures, Modeling, and Simulation, 2008. SAMOS 2008. International Conference on. IEEE, 2008.
497
