cISP: A Speed-of-Light Internet Service Provider

Main Problem

• Main problem: reduce latencies over the Internet to the lowest possible.

• Elaborations: user experience in many interactive network applications depends crucially on achieving low latency; a seemingly small increases in latency can negatively impact user experience can negatively impact user experience, and subsequently, revenue for service providers. Further, wide-area latency is often the bottleneck, and the Internet is not designed to treat low latency as a primary objective.

Main Insight

• The paper explores the design of an Internet Service Provider that provides nearly speed-of-light latency by exploiting wireless electromagnetic transmissions, which can be realized with point-to-point microwave antennas mounted on towers.

• The paper proposes a hybrid design that augments the Internet's fiber connectivity with nearly straight-line wireless links; the low-latency links are used judiciously where they provide the maximum latency benefits, and only for the high-impact but small proportion, in terms of bytes, of internet traffic that is latency-sensitive

Key Strength

• Practical problem to address, a series of killer applications is discussed. Some applications focused on user interactivity like the augmented and virtual reality, tele-presence and tele-surgery, musical collaboration over long-distances, etc., can all benefit from low latency connectivity. Less user-centric applications like real-time bidding for web page advertisements and block propagation in blockchains would also benefit. In section 7, the paper studies cISP's benefits for two example applications.

• The paper identifies numerous practical challenges being network design and routing for deploying cISP in real life and provides comprehensive measurements and evaluations. This includes studies of the impact of weather on performance, tower height and availability, and integration into the Internet.

• The paper proposes methodology to solve a large-scale cost-optimal network design problem, and makes the approach fast to solve practically.

Key Weakness

• Further research and actual deployments may be necessary for fully addressing the effectiveness and access whether cISP is practical in real life

• Some discussions on

  • More ethical way to allocate resources

  • Mean stretch as the main metric [adding more towers after a certain point doesn't help connect new areas]

    • Maybe for video streaming applications people care about 99 percentile as metric

  • Cost discussion: $0.81/GB, large or reasonable

  • Trading data evaluation generalize to national network?

  • Changing weather patterns impact the reliability of free-space network?

  • Limitations of the approach, last-mile communication, hybrid design (fiber / microwave)