Zafar Ayyub Qazi's Homepage
a picture of zafar qazi


Zafar Ayyub Qazi

Assistant Professor at LUMS

Email: zafar.qazi [at] lums [dot] edu [dot] pk

Here is my CV

Home | News | Research | Publications | Teaching | Services | Students | Resources


About

I am an Assistant Professor in the Computer Science Department at LUMS.
My research aims to improve the user experience, affordability, and privacy of networked systems.
My current research centers on two broad themes: (1) Rethinking the architecture and design of next-generation cellular networks to enable emerging real-time applications,
(2) Rethinking the design of the Web for improving affordability, privacy, and user experience.
Formerly, I was a Postdoctoral scholar in the EECS Department at UC Berkeley. Before that, I earned my PhD at State University of New York, Stony Brook.

News

  • 10/21: My PhD student Mukhtiar Ahmed has won the SBARA research award at LUMS. Congratulations Mukhtiar!
  • 08/21: Our proposal for rethinking Web for affordability and inclusion accepted in HotNets 2021.
  • 08/21: Had the opportunity to serve in some great TPCs this year: SIGCOMM, HotNets, and SoCC.
  • 04/21: Recognized by AI 2000 in their list of most influential scholars for the last decade. Feel fortunate to have worked with some wonderful collaborators.
  • 05/20: Our work on the design of a new control plane for cellular networks accepted in ACM SIGCOMM 2020. Congratulations to Neutrino team at LUMS!
  • 01/20: Our work deconstructing Google's Web Light Service to appear in WWW 2020.
  • 09/19: LUMS will be hosting the National Center for Big Data and Cloud Computing . I will be leading the 5G efforts in the Distributed Computing Lab.
  • 06/18: Won Google Faculty Research Award.

    Selected Research

      A Low Latency and Consistent Cellular Control Plane [SIGCOMM'20]

      Neutrino


      5G networks aim to provide ultra-low latency and higher reliability to support emerging and near real-time applications such as augmented and virtual reality, remote surgery, self- driving cars and multi-player online gaming. This imposes new requirements on the design of cellular core networks. A key component of the cellular core is the control plane. Time to complete control plane operations (e.g. mobility handoff, service establishment) directly impacts the delay experienced by end-user applications. In this work, we design Neutrino, a cellular control plane that provides an abstraction of low latency and reliable access to user state. Our testbed evaluations based on real cellular control traffic traces show Neutrino provides an improvement in control procedure completion times by upto 3.1x without failures, and upto 5.6x under control plane failures, over existing cellular core proposals. We also show how these improvements translate into improving end-user application performance; for AR/VR applications and self-driving cars, Neutrino performs 2.5x and 2.8x better, respectively, as compared to existing EPC.

      Mukhtiar Ahmad, Syed Usman Jafri, Azam Ikram, Wasiq Qasmi, Ali Nawazish, Zartash Uzmi, Zafar Ayyub Qazi. A Low Latency and Consistent Cellular Control Plane. ACM SIGCOMM 2020


      A High Performance Packet Core for Next Generation Cellular Networks [SIGCOMM'17]

      PEPC


      Cellular traffic continues to grow rapidly making the scalability of the cellular infrastructure a critical issue. However existing EPC solutions scale poorly, in particular with increasing numbers of devices or signaling traffic. The key reason for this poor scalability is the manner in which EPC functions are decomposed which leads to device state being duplicated across multiple components, which in turn results in frequent interactions between these components. To address these limitations, we proposed PEPC, a new state-driven system architecture in which we slice EPC by user such that state for single user is consolidated in one location and EPC functions are (re)organized for efficient access to this consolidated state. We show that PEPC achieves 3-7X higher throughput than comparable software EPCs implemented in industry and enables efficient state migration and customization of processing pipelines.

      Zafar Ayyub Qazi, Melvin Walls, Aurojit Panda, Vyas Sekar, Sylvia Ratnasamy, Scott Shenker. A High Performance Packet Core for Next Generation Cellular Networks. ACM SIGCOMM 2017


      All-Wireless Inter-Rack Data Center Design Using FSOs [SIGCOMM'14]

      Firefly


      Designing data centers (DC) is hard. A robust data center network must satisfy several goals: high throughput, low equipment and management cost, robustness to dynamic traffic patterns, incremental expandability, low cabling complexity, and low power and cooling costs. We explored the vision of a flexible, all-wireless, inter-rack datacenter (DC) fabric. Towards this end, we proposed FireFly, which is a flexible, all-wireless, inter-rack datacenter fabric that uses free-space optics and achieves performance close to a full bisection bandwidth network but at only 40-60% of the cost.

      Navid Hamed Azimi, Zafar Ayyub Qazi, Himanshu Gupta, Vyas Sekar, Samir Das, Himanshu Shah, Ashish Tanwer. FireFly: A Reconfigurable Wireless Datacenter Fabric, using Free-Space Optics. ACM SIGCOMM 2014.


      Simplifying Middlebox Policy Enforcement using SDN [SIGCOMM'13]

      SIMPLE


      Networks today rely on middleboxes (e.g., firewalls, VPN gateways, proxies, intrusion detection and prevention systems, WAN optimizers) to provide critical performance, security, and policy compliance capabilities. Achieving the performance and security benefits that middleboxes offer, however, is highly complex. In this project, we addressed algorithmic and system design challenges to demonstrate the feasibility of using SDN to simplify middlebox management. We propose SIMPLE, an orchestration layer for managing middleboxes using SDN. It ensures middlebox-specific policy enforcement and improves middlebox loading balancing by 6X compared to today's deployments.

      Zafar Ayyub Qazi, Cheng-chun Tu, Luis Chiang, Rui Miao, Vyas Sekar, Minlan Yu. SIMPLE-fying Middlebox Policy Enforcement Using SDN. ACM SIGCOMM 2013




    Publications


    Teaching

  • Mukhtiar Ahmad (PhD Student: 2018 - Present )
  • Syed Ali Nawazish (Research Associate) (Joining University of Utah for PhD in Fall 2022)
  • Muhammad Tahir (Research Associate, co-advised with Prof. Ihsan Qazi)
  • Taimoor Tariq (Research Assistant) (Joining UIUC for PhD in Fall 2022)
  • Talha Waheed (Research Assistant, Co-advised with Prof. Ihsan Qazi) (Joining UIUC for PhD in Fall 2022)
  • Muhammad Basit Iqbal Awan (Research Assistant) (Joining University of Utah for PhD in Fall 2022)
  • Rumaisa Habib (Undergrad RA, co-advised with Prof. Ihsan Qazi)
  • Sarah Tanveer (Undergrad RA, co-advised with Prof. Ihsan Qazi)
  • Faaiq Bilal (Undergrad RA)
  • Aimen Inam (MS, RA)
  • Abdullah Khan (Undergrad RA)
  • Haseeb Ahmed (Undergrad RA, co-advised with Prof. Ihsan Qazi)
  • Mutahar Ali (Undergrad RA, co-advised with Prof. Ihsan Qazi and Prof. Agha Ali Raza)
  • Fatima Sohail (Undergrad RA, co-advised with Prof. Ihsan Qazi and Prof. Agha Ali Raza)

    Alumni

  • Syed Usman Jafri (PhD Student at Purdue University)
  • Ammar Tahir (PhD Student at UIUC)
  • Maleeha Masood (PhD Student at UIUC)
  • Azam Ikram (PhD Student at Purdue University)
  • Muhammad Abdullah (PhD Student at EPFL)
  • Arsalan Jumani (Senior Software Engineer at Apple)
  • Abdul Manan (PhD Student at Brown University)
  • Ahmed Hassan (PhD Student at Minnesota University)
  • Ghulam Murtaza (PhD Student at Brown University)
  • Ahmed Faraz (PhD Student at Virginia Tech)
  • Abrar Tariq (PhD Student at University of Illinois)
  • Ehsan Latif (PhD Student at University of Georgia)
  • Fizza Zafar (MS ETH Zurich)
  • Salman Munaf (MS Wisconsin Madison)
  • Osama Khurshid (MS Georgia Institue of Technology)

    Resources

  • [A collection of advices for junior faculty] by Jean Yang
  • [Advice] I got when I started my Ph.D.