Hasil Pencarian  ::  Simpan CSV :: Kembali

Abstrak :
Penggunaan Load Balancing pada aplikasi E-Voting memberikan hasil performa yang effisien untuk hasil test GET (download) dan hasil yang tidak effisien untuk hasil test POST (upload). Dengan menggunakan Load Balancing, angka yang diperoleh untuk throughput sebesar 750kbps. Pada test POST membutuhkan proses read dan write, throughput yang diperoleh kurang dari 2kbps. Pengukuran POST mendapatkan hasil yang tidak efisien dikarenakan penggunaan content server secara virtual (3 server pada satu machine). Algoritma Least Current Request menunjukkan performa terbaik dibandingkan dengan 3 algoritma lainnya, yaitu Least Response Time, Weighted Round Robin, dan Weighted Total Traffic. Kinerja Load Balancing tidak hanya dipengaruhi oleh jumlah paket data yang diakses, tetapi juga proses request apa yang dilakukan. Solusi terbaik adalah, Load Balancing diimplementasikan dengan menggunakan mesin yang memiliki spesifikasi yang tinggi pada database server dan setiap server pada server farms tidak dijalankan secara virtual. ......The use of Load Balancing in the E-Voting application gives an efficient result for the GET (download) test, and an inefficient one for the POST (upload) test. Using Load Balancing, the resulting throughput is 750 kbps. The POST test needs read and write processes, making the resulting throughput less than 2 kbps. The POST test obtained inefficient results because the uses of content server virtualy (3 servers in one machine). Least Current Request algorithm shows the best performance compared to the other three in the measurement, which are Least Response Time, Weighted Round Robin, and Weighted Total Traffic. The performance of Load Balancing not only influenced by the numbers of data packet accessed, but also by the request process. The best solution is to implement Load Balancing on a high specification system on the server database, with no virtual machines for the servers

Abstrak :
In recent years, there is a significant increase on the demand of multimedia services. The use and delivery of various multimedia applications, collectively and continuously, creates increasing demand on network bandwidth. However, in most of the countries, the available bandwidth capacity remains very limited Majority users living in these regions are having difficulties to access such applications. A solution to this problem is by embedding the capability to dynamically change the format of bandwidth-intensive multimedia applications to a more efficient one into the network. The process which alters multimedia application from one format to another format is known as Transcoding. For example, cinepak format requiring high bandwidth is transcoded into lower format such as example H263. Since transcoding process is very demanding in terms computing resources, load balancing strategy in a newark environment heaps to share the load among the transcoder processor; and in turn, improves throughput and efficiency. This paper describes the implementation and analysis of applying transcoder with load balancing mechanism in a video streaming environment. Three load balancing algorithms are considered and real tests using various video and audio formats are performed in this environment. Eventhough the load balancing strategy improves the performance of the system, we still find that as the number processes increases the ability to deliver multimedia applications through transcoding process decreases. Moreover, we also conclude that our system still poses scalability problems and therefore further work is required.

Abstrak :
[ABSTRAK
VANET adalah sebuah jaringan tanpa infrastruktur yang terdiri dari entitas atau node bergerak yang memiliki perangkat wireless untuk berkomunikasi satu sama lain. VANET memiliki topologi yang dinamis yang disebabkan oleh mobilitas dari node-node pada VANET. Dalam simulasi pada vanet dibutuhkan model mobilitas yang mencerminkan keadaan sebenarnya. Model mobilitas menentukan pergerakan dari entitas/node, bagaimana node bergerak, kecepatan dan percepatan dalam VANET. Untuk mencegah kemacetan di jalan raya dilakukan load balancing/pengalihan arus kendaraan untuk membagi trafik kendaraan, dengan satu jalur jalan raya, kepadatan kendaraan akan sangat tinggi, pada topologi grid map dengan load balancing, kepadatan kendaraan dapat berkurang. Pada simulasi didapatkan penurunan kinerja pada model mobilitas grid map dengan load balancing. Pada load balancing, perubahan topologi yang menjadi lebih besar mengakibatkan penurunan kinerja dari load balacing pada model mobilitas gridmap dengan 19,3% pada delay, 5,08% pada packet delivery ratio dan 5,46% pada throughput dalam paramater AODV dan 0,01% pada delay, 12,70% packet delivery ratio dan 12,73% pada throughput dalam parameter DSDV.

---

ABSTRACT
VANET is an infrastucture less network that consist of entities or mobile nodes that contain wireless device for communicating each other. VANET has dynamic topology due to node mobility. VANET simulation is need mobility model that reflecting the real situation. Mobility model determines the movement of mobile nodes how their location, velocity and acceleration in VANET. To prevent congestion on the road, grip map topology is use to divide the traffic, with the one lane road map, the vehicles density will very high, with load balance grid map topology the vehicles density will separated to other road. On simulation result we get the decrease performance on the load balacing of grid map mobility model. On the load balancing dynamic topology will cause scalability of the network and impact to performance of mobility model. From simulation, load balancing of grid map mobility model has lower performace than common grid map with 19,3% on delay, 5,08% on packet delivery ratio, and 5,46% on throughput with AODV parameters and 0,01% on delay, 12,70% on packet delivery ratio and 12,73% on throughput with DSDV parameters;VANET is an infrastucture less network that consist of entities or mobile nodes that contain wireless device for communicating each other. VANET has dynamic topology due to node mobility. VANET simulation is need mobility model that reflecting the real situation. Mobility model determines the movement of mobile nodes how their location, velocity and acceleration in VANET. To prevent congestion on the road, grip map topology is use to divide the traffic, with the one lane road map, the vehicles density will very high, with load balance grid map topology the vehicles density will separated to other road. On simulation result we get the decrease performance on the load balacing of grid map mobility model. On the load balancing dynamic topology will cause scalability of the network and impact to performance of mobility model. From simulation, load balancing of grid map mobility model has lower performace than common grid map with 19,3% on delay, 5,08% on packet delivery ratio, and 5,46% on throughput with AODV parameters and 0,01% on delay, 12,70% on packet delivery ratio and 12,73% on throughput with DSDV parameters, VANET is an infrastucture less network that consist of entities or mobile nodes that contain wireless device for communicating each other. VANET has dynamic topology due to node mobility. VANET simulation is need mobility model that reflecting the real situation. Mobility model determines the movement of mobile nodes how their location, velocity and acceleration in VANET. To prevent congestion on the road, grip map topology is use to divide the traffic, with the one lane road map, the vehicles density will very high, with load balance grid map topology the vehicles density will separated to other road. On simulation result we get the decrease performance on the load balacing of grid map mobility model. On the load balancing dynamic topology will cause scalability of the network and impact to performance of mobility model. From simulation, load balancing of grid map mobility model has lower performace than common grid map with 19,3% on delay, 5,08% on packet delivery ratio, and 5,46% on throughput with AODV parameters and 0,01% on delay, 12,70% on packet delivery ratio and 12,73% on throughput with DSDV parameters]

Abstrak :
Pertumbuhan traffic yang terjadi menyebabkan Capital Expenditure (CAPEX) yang harus dikeluarkan oleh operator seluler setiap tahun semakin meningkat dan kompleksitas operasional jaringan semakin tinggi sehingga perlu peningkatan efisiensi jaringan salah satunya melalui sistem otomatisasi menggunakan Self-Organizing Network (SON). Adanya penerapan salah satu use case Self-Organizing Network (SON) yaitu Mobility Load Balancing (MLB) dapat meningkatkan efisiensi Capital Expenditure (CAPEX) yang harus dikeluarkan oleh operator seluler karena adanya proses load transfer antara cell secara otomatis sehingga ekspansi yang dikarenakan isu keterbatasan kapasitas cell dapat dikurangi jumlahnya. Penelitian tesis ini dilakukan untuk membuat pemodelan dan analisis implementasi dari Mobility Load Balancing (MLB) di jaringan 3G. Model ini dapat menentukan algoritma bekerjanya Mobility Load Balancing (MLB) dan nilai investasi maksimum bulanan yang layak dikeluarkan oleh operator seluler terkait implementasi Mobility Load Balancing (MLB). ......The growing of cellular traffic nowadays causing the increment of Capital Expenditure (CAPEX) to be incurred by the mobile operators annually and the increment of network operation complexity so that the increment of network efficiency is needed through the automation system using Self-Organizing Network (SON). The implementation of one of Self-Organizing Network (SON) use case, namely Mobility Load Balancing (MLB), can improve the efficiency of Capital Expenditure (CAPEX) due to the transfer of load between the cells automatically so that the cell expansion number due to capacity issue can be reduced. This thesis research was carried out to make the modeling and analysis of the implementation of the Mobility Load Balancing (MLB) in the 3G network. This model can determine the algorithm of Mobility Load Balancing (MLB) and the monthly maximum eligible investments issued by mobile operators regarding the implementation of Mobility Load Balancing (MLB).

Abstrak :

ABSTRAK
Pada keadaan nyata, beban suatu node pada j3ringan sangat fluktuatif, bergantung pada banyaknya pemakai mesin tersebut saat flu. Beban pemakaian ini sangat berpengaruh pada kecepatan pemrosesan mesin tersebut.

Hal ini menyulitban bagi pemrograman aplikasi paralel, karena aplikasi paralel akan efisien bila aphkasi dapat membagi beban pemrosesan (worklow secara proporsional. Karena kecepatan pemrosesan mesin berubah-ubah menurut beban pemakaian mesin tersebut, maka pedu dicari metoda untuk menilai untuk kerja tiap mesin pada jaringan secara dinamis.

Pada proses pengerjaan skripsi ini, dikembangkan metoda untuk menilai unjuk kerja dinamis dari mesin dengan menggunakan parameter waktu interval antar checkpointing. Metoda tersebut diterapkan pada aplikasi dengan paradigma SIMD, dimana data yang diproses berjumlah besar, seragam dan pemrosesannya menggunakan fungsi yang sama dan berulang.

Pertama-tama dikembangkan metoda check pointing dan migrasi proses dengan teknik high-level. Segala aspek dan latar belakang dari rancangan higb-level checkpointing dan migrasi proses akan dikupas lebih ]anjut.

Selanjutnya dilakukan pengujian apakah waktu interval antar checkpointing memenuhi syarat untuk digunakan sebagai parameter untuk menilai unjuk kerja mesin secara dinamis. Akan diuji apakah parameter tersebut mencerminkan kecepatan proses dari mesin-mesin yang digunakan.

Yang terakhir, dengan menggunakan parameter tersebut dan metoda migrasi proses, dikembangkan algoritma untuk melakukan dynamic loud balancing. Peningkatan speed-up aplikasi paralel akan didapatkan melalui pembagian beban pemrosesan yang lebih proporsional.

Abstrak :
This paper explains the development of a parallel computing platform using the available LAN at the Inter University Center for Computer Science of Indonesia. A software toll like parallel virtual machine enables the networked stations to be harnessed as a huge computing resource, combining various type of computing nodes to run parallel computing tasks. The research was aimed at first: to develop a number of parralel high complexity algorithms such as modelling air pollution distribution based on Gaussian Air Pollution Dispersion Formula which simulates dispersion of pollutant emiotted from industrial exhaust and also from vehicle exhaust ont the road. Secondly these algorithms were used to test the platform to obtain networkd behaviour in supporting the implementation of parallel algorithms. From the experiments conducted there are strong indications that the network based computing facility could be used as a suitable platform for running parallel programs with some restrictions. Optimum performance with near linear speed up could be obtained by carefully partitioning the problems, limiting message passing activities and balancing the load of the participating processors.