By Rob Knies, Managing Editor, Microsoft Research
We’ve been spoiled. In this age of virtually instantaneous communication, we have developed a thirst for immediate access to information. Waiting is a pain in the Internet age. Soon is not sufficient; we need it now.
Researchers from Microsoft Research Redmond (opens in new tab) are doing their best to comply.
Spotlight: blog post
GraphRAG auto-tuning provides rapid adaptation to new domains
GraphRAG uses LLM-generated knowledge graphs to substantially improve complex Q&A over retrieval-augmented generation (RAG). Discover automatic tuning of GraphRAG for new datasets, making it more accurate and relevant.
A Case for Adapting Channel Width in Wireless Networks (opens in new tab), co-written by Ranveer Chandra (opens in new tab), Ratul Mahajan (opens in new tab), Thomas Moscibroda (opens in new tab), and Paramvir Bahl of Microsoft Research Redmond, in collaboration with Ramya Raghavendra of the University of California, Santa Barbara, is one of eight papers from Microsoft Research that has been accepted for SIGCOMM (opens in new tab), the Association for Computing Machinery’s annual conference of the Special Interest Group on Data Communications.
A total of 36 papers were accepted for SIGCOMM 2008, to be held in Seattle from Aug. 17 to 22, meaning that 22 percent of the accepted papers for the conference come from Microsoft Research. Four of the organization’s six labs are represented; in addition to the Redmond lab, Microsoft Research Cambridge (opens in new tab), Microsoft Research Asia (opens in new tab), and Microsoft Research Silicon Valley also have papers to be presented during the event. Six of Microsoft Research’s eight papers were collaborative efforts with university partners.
In addition, Microsoft Research scientists are serving in key roles as SIGCOMM 2008 organizers. Bahl, a principal researcher in Redmond’s Networking Research Group (opens in new tab), is one of three general chairs for the conference. Mahajan, a researcher in the same group, is acting as the local-arrangements chair, and he and Albert Greenberg, also a principal researcher in the Networking Research Group, are on the SIGCOMM Program Committee.
In the aforementioned paper, the authors argue that wireless systems that enable variable channel widths—the width of the spectrum over which transmitters spread their signals—can deliver significant benefits.
Channel width can affect throughput rates, range, and power consumption. The researchers have determined that when the need for throughput is low, a narrower channel can increase range and, at the same time, reduce power consumption. In traditional, fixed-width systems, range and power consumption are conflicting values.
“The unlicensed spectrum is overcrowded with a large number of wireless devices—Wi-Fi laptops, Bluetooth headsets, baby monitors, cordless phones, and many others,” Chandra explains. “To achieve the best performance of wireless networks in this setting, it is necessary to make the wireless networks ‘spectrum aware.’ They should operate in the cleanest frequency spectrum. Furthermore, the network should operate on as much of the spectrum that is available. Hence they should not only adapt the channel number (center frequency), but also the channel width.
“In this work, we have shown the feasibility of adapting the channel width—a fundamental, yet unexplored parameter—from software on off-the-shelf Wi-Fi cards. This is also the first paper to characterize the properties of different channel widths and to show that adaptive-channel-width systems significantly improve the performance of wireless networks.”
Another paper accepted for SIGCOMM 2008 is Spamming Botnets: Signatures and Characteristics (opens in new tab), written by Yinglian Xie, Fang Yu, Kannan Achan, and Rina Panigrahy of Microsoft Research Silicon Valley, along with Microsoft colleagues Geoff Hulten and Ivan Osipkov.
The paper, an extension of the work Xie and Yu have been pursuing to combat the use of zombie botnets that distribute spam online. The paper outlines a system to characterize and battle spamming botnets by analysis of spam payload and spam-server traffic properties.
Papers from Microsoft Research accepted for SIGCOMM 2008:
A Case for Adapting Channel Width in Wireless Networks (opens in new tab)
Ranveer Chandra, Microsoft Research Redmond; Ratul Mahajan,Microsoft Research Redmond; Thomas Moscibroda, Microsoft Research Redmond; Ramya Raghavendra, University of California, Santa Barbara); and Paramvir Bahl, Microsoft Research Redmond
DCell: A Scalable and Fault-Tolerant Network Structure for Data Centers (opens in new tab)
Chuanxiong Guo, Microsoft Research Asia; Haitao Wu, Microsoft Research Asia; Kun Tan, Microsoft Research Asia; Lei Shi, Tsinghua University; Yongguang Zhang, Microsoft Research Asia; and Songwu Lu, UCLA
Donnybrook: Enabling Large-Scale, High-Speed, Peer-to-Peer Games (opens in new tab)
Ashwin Bharambe, Carnegie Mellon University; John R. Douceur, Microsoft Research Redmond; Jacob R. Lorch, Microsoft Research Redmond; Thomas Moscibroda, Microsoft Research Redmond; Jeffrey Pang, Carnegie Mellon University; Srinivasan Seshan, Carnegie Mellon University; and Xinyu Zhuang, Carnegie Mellon University
Interactive WiFi Connectivity for Moving Vehicles (opens in new tab)
Aruna Balasubramanian, University of Massachussetts Amherst; Ratul Mahajan, Microsoft Research Redmond; Arun Venkataramani, University of Massachussetts; Amherst Brian Neil Levine,University of Massachussetts Amherst; and John Zahorjan, University Of Washington
Network Exception Handlers: Host-Network Control in Enterprise Networks (opens in new tab)
Thomas Karagiannis, Microsoft Research Cambridge; Richard Mortier, Vipadia Ltd; and Antony Rowstron, Microsoft Research Cambridge
Predictable Performance Optimization for Wireless Networks (opens in new tab)
Yi Li,University of Texas at Austin; Lili Qiu,University of Texas at Austin; Yin Zhang, University of Texas at Austin; Ratul Mahajan, Microsoft Research Redmond; and Eric Rozner, University of Texas at Austin
Spamming Botnets: Signatures and Characteristics (opens in new tab)
Yinglian Xie, Microsoft Research Silicon Valley; Fang Yu, Microsoft Research Silicon Valley; Kannan Achan, Microsoft Research Silicon Valley; Rina Panigrahy, Microsoft Research Silicon Valley; Geoff Hulten, Microsoft; and Ivan Osipkov, Microsoft
What’s Going On? Extracting Communication Rules In Edge Networks (opens in new tab)
Srikanth Kandula, Massachusetts Institute of Technology; Ranveer Chandra, Microsoft Research Redmond; and Dina Katabi, Massachusetts Institute of Technology
Systems and networking