Hasil Pencarian  ::  Simpan CSV :: Kembali

"Models that include a notion of time are ubiquitous in disciplines such as the natural sciences, engineering, philosophy, and linguistics, but in computing the abstractions provided by the traditional models are problematic and the discipline has spawned many novel models. This book is a systematic thorough presentation of the results of several decades of research on developing, analyzing, and applying time models to computing and engineering."

"Aplikasi berbasis jaringan internernet real-time seperti VoIP dan Video Conferencing sangat sensitif terhadap gangguan berupa ketersediaan bandwidth, terjadinya delay, packet loss, dan jitter. Untuk jaringan data dengan trafik bervariasi (seperti voice, video, data), dibutuhkan perlakuan khusus terhadap trafik data tertentu. Implementasi teknologi Quality of Service u n t u k menguji kinerja umum dilakukan dan teknik yang paling populer adalah Differentiated of Service (DiffServ). Namun dengan meningkatnya penetrasi trafik data pada jaringan komunikasi, masalah skalabilitas dan efisiensi menjadi penting. Multi- Protocol Label Switching (MPLS) memberikan kecepatan, skalabilitas serta efisiensi terhadap unjuk kerja jaringan. Dengan mengintegrasikan kedua teknologi tersebut, diharapkan unjuk kerja jaringan semakin baik bagi aplikasi atau layanan yang memerlukan garansi Quality of Service (QoS). Dalam penelitian ini implementasi teknologi MPLS DiffServ-Aware tidak hanya dilakukan pada jaringan IPv4 namun juga pada jaringan IPv6.Simulasi menunjukkan aplikasi yang sensitif terhadap QoS seperti VoIP, implementasi MPLS/DiffServ mampu memberikan garansi QoS dengan delay mencapai 87 ms, packet loss sebesar 0% dan jitter yang sangat kecil. Untuk aplikasi video conferencing, unjuk kerja juga lebih baik jika dibandingkan dengan hanya menggunakan teknologi IP Best- Effort, DiffServ, maupun MPLS secara independen. Untuk jaringan IPv6, unjuk kerja throughput lebih baik dari IPv4, namun secara keseluruhan unjuk kerja jaringan IPv4 masih lebih baik dibandingkan jaringan IPv6.

---

Real-time network based applications such as VoIP and Video Conferencing are critical when it deals with bandwidth, delay, packet loss, and jitter. In a service integrated network (such as voice,video, and data services), it is needed to treat a service accordingly. The implementation of Quality of Service are common in today's networks : one of those technologies is Differentiated of Service (DiffServ). Nevertheless, with higher penetration of data traffics from time to time, the issues of scalability and eficiency appear significantly. Multi-Protocol Label Switching (MPLS) offers a good solution for improving the network performance. By integrating those two technologies, it is expected to improve the overall network performance, especially for service or application that needs QoS-guaranteed. This research not only focus on the implementation of MPLS DiffServ- Aware in IPv4 networks environment but also in IPv6 as well.Simulation results, showed that MPLS/DiffServ technology provides best performance for real-time or QoS-sensitive applications. For VoIP application, the delay is recorded as 87 ms, packet loss 0% and a very low jitter. For video conferencing service, those QoS parameters are also better if compared to those technologies if are implemented separately. Althought IPv6 showed a higher throughput performance, still IPv4 showed a better performance generally."

"Aplikasi kolaboratif berbasis real-time menjadi bagian dari kehidupan sehari-hari manusia, terutama di masa pandemi ketika setiap orang sulit untuk bertemu secara langsung. Salah satu aplikasi tersebut adalah aplikasi collaborative whiteboard, yang bisa digunakan untuk menggambar secara bebas. Salah satu implementasi dari sistem collaborative whiteboard adalah Tldraw. Tldraw yang semula dikembangkan menggunakan arsitektur client-server kemudian diimprovisasi menggunakan arsitektur peer-to-peer yang memanfaatkan teknologi WebRTC, menghasilkan aplikasi Tldraw yang rendah latensi. Namun, aplikasi Tldraw tersebut dinilai tidak bersifat scalable, karena arsitektur peer-to-peer tersebut menggunakan topologi full-mesh. Dilatarbelakangi oleh hal tersebut, penelitian ini mengusulkan optimasi aplikasi Tldraw menggunakan topologi random peer sampling. Usulan optimasi tersebut dinilai memiliki pertumbuhan penggunaan sumber daya yang lebih landai apabila dibandingkan dengan Tldraw P2p, meskipun memiliki sedikit overhead.

---

A real-time based collaborative applications are taking part in human daily life, especially in the pandemic era when it is prohibited to directly meet other people. One example of those kind of application is collaborative whiteboard application, which can be use to draw things freely. Tldraw previously developed using client-server architecture, which later improved using peer-to-peer architecture using WebRTC technology, resulting a low-latency Tldraw application. Nevertheless, those improvisations seem to be not scalable, since the peer-to-peer architecture use full-mesh network topology. Motivated by that, this research propose optimization for Tldraw application using random peer sampling topology. Those optimizations resulting in lower growth of cpu, memory, and bandwidth usage with little overhead."