Algorithms and Protocols for Next Generation WiFi Networks
MetadataShow full item record
The Internet traffic requirement has grown remarkably in the last couple of decades, and it is predicted to scale even further in the upcoming years. On the other hand, Internet users are increasingly relying on WiFi for the last mile connectivity. Thus, technologies that can fundamentally enhance WiFi are desirable. Research on next generation WiFi networks is focusing on a variety of goals including performance (i.e., system throughput and channel efficiency), controllability, security, power consumption, adaptability, etc. The goals of our research are improving performance to support the growth of requirements, and promoting controllability to support service differentiation and even performance prediction. Performance goals pertain to achieving better throughput and channel efficiency for a wide range of network conditions with the highest degree of adaptability. Controllability goals are akin to the goals of software-defined networks and focus on future-proofing systems, manageability, and service differentiation capabilities. First, we propose algorithms that improve the performance in the physical (PHY) layer with only changes in Access Points (APs). Smart antenna technology plays an important role for performance enhancement in the PHY layer of WiFi. While most smart antenna techniques require changes in both APs and stations (STAs), beamforming is a mechanism that can be applied with only changes in APs. We propose FastBeam, a set of algorithms that can provide benefits of beamforming to legacy nodes by only adopting new APs. Second, we consider future-proofing networks with a central controller and enable micro-level controllability to support service differentiation and performance prediction. We focus on how to enable the controllability of WiFi networks without compromising their scalability when a central controller is available. We introduce a media access control (MAC) protocol called Rhythm, which transfers the control of WiFi networks into centralized scheduling, with the properties of (i) low protocol overhead, (ii) work conservation in the presence of non-backlogged nodes, (iii) robustness to partial connectivity scenarios. Specifically, Rhythm furnishes all nodes in WiFi networks with a target schedule that has been determined by a central controller. The nodes in WiFi networks then operate in a purely distributed fashion to meet the target schedule. We refer to such a system behavior as “scheduled WiFi.” Finally, we consider the backward compatibility issues of scheduled WiFi. Though Rhythm brings micro-level controllability to WiFi, it can only operate in an ideal environment, where i) no other legacy nodes are present and ii) the network topology is known. To provide better backward compatibility for implementing scheduled WiFi, we propose a new MAC protocol, LWT. LWT not only achieves scheduled WiFi in a purely distributed fashion, but is also more friendly to legacy nodes, and does not require the network topology information. LWT uses a novel mechanism, called Switch, to deal with hidden terminal problems. Switch provides a way to utilize overlay control channels better in WiFi networks. In addition to the details of how Switch is used in LWT, we further explored how Switch can be used in other situations, such as i) extending the range of carrier sense, ii) early collision termination, and iii) improving the efficiency of WiFi backoff mechanism.