In this paper, we investigate the optimal line-of-sight (LoS) coverage problem for multiple access point (multi-AP) mmWave wireless LANs in indoor scenarios. Due to the weak diffraction ability of mmWave signals at 60 GHz, maintaining LoS communications between APs and client devices is critical to achieve ultra-high data rates with mmWave communications. We focus on the use of multiple APs deployed to maximize LoS coverage in a target area, and we develop multi-AP placements that maximize LoS coverage by means of both analytical and algorithmic methods. We consider two main scenarios, which differ in their assumptions about knowledge of obstacles and clients. In a random-obstacle, random-client scenario, we derive the LoS-optimal positions of APs by solving a thinnest covering problem. For a fixed-obstacle, random-client scenario, we propose an efficient algorithm that produces a multi-AP placement, which is shown through simulation to provide near-optimal LoS coverage. Finally, through extensive ns-3 simulations based on the IEEE 802.11ad protocol and mmWave-specific channel models, we show that our multi-AP placements are significantly better than existing placement approaches, both in terms of LoS coverage and aggregate throughput.

mmWave is emerging as an essential technology for next-generation wireless networks due to its capability of delivering multi-gigabit throughput performance. To achieve such a promising performance in mmWave communications, Line-of-sight (LOS) connectivity is a critical requirement. In this work, we explore the strategy of infrastructure mobility to alter the location of an access point (AP) in order to provide LOS connectivity to stations (STAs) in indoor mmWave WiFi networks. Through both simulation-based and theoretical analyses, we make a detailed case for infrastructure mobility by identifying the impact of AP mobile platforms configurations on network performance and propose a ceiling-mounted mobile (CMM) AP model. Then, we compare the performance of a CMM AP with multiple static APs, and we identify that the throughput and fairness performance of a CMM AP is better than as many as 5 ceiling-mounted static APs.

Millimeter-wave communication is a highly promising technology to deliver multi-gigabit-per-second transmission rates for next-generation wireless LANs (WLANs). To achieve such ultra-high throughput performance in indoor scenarios, line-of-sight (LoS) connectivity becomes a critical requirement. Prior work has proposed access point (AP) mobility as an approach to improve LoS conditions and, thereby, approach optimum mmWave WLAN performance. In this work, we present a comprehensive simulation study of linear AP mobility that investigates various dimensions, including the number of mobile APs, the placement of the mobile AP platforms, and the length of the platforms. The results show how WLAN performance varies across these dimensions and also compares the results against a varying number of static APs to quantity the performance gains achievable from mobility. The results show that even 2 or 3 mobile APs can significantly outperform a much larger number of static APs and that deploying up to 3 mobile APs in a room brings substantial performance gains.

Line-of-sight (LOS) is a critical requirement for mmWave communication. In this work, we make the case for a ceilingmounted mobile (CMM) AP by comparing its performance with other types of AP mobility and single static AP. We then present Hawkeye to solve the optimal location discovery problem for a CMM AP using a machine learning (ML) algorithm. Hawkeye relies purely on the connectivity matrix between STAs and the AP to decide if and where the AP should move to for maximizing LOS connectivity. Using a prototype implementation, we show that the throughput of Hawkeye is 219% and 129% compared with single static AP and other approaches for AP mobility, respectively.

mmWave communication in 60GHz band has been recognized as an emerging technology to support various bandwidth-hungry applications in indoor scenarios. To maintain ultra-high throughputs while addressing potential blockage problems for mmWave signals, maintaining line-of-sight (LoS) communications between client devices and access points (APs) is critical. To maximize LoS communications, one approach is to deploy multiple APs in the same room. In this paper, we investigate the optimal placement of multiple APs using both analytical methods and simulations. Considering the uncertainty of obstacles and clients, we focus on two typical indoor settings: random-obstacle-random-client (RORC) scenarios and fixed-obstacle-random-client (FORC) scenarios. In the first case, we analytically derive the optimal positions of APs by solving a thinnest covering problem. This analytical result is used to show that deploying up to 5 APs in a specific room brings substantial performance gains. For the FORC scenario, we propose the shadowing-elimination search (SES) algorithm based on an analytic model to efficiently determine the placement of APs. We show, through simulations, that with only a few APs, the network can achieve blockage-free operation in the presence of multiple obstacles and also demonstrate that the algorithm produces near-optimal deployments. Finally, we perform ns-3 simulations based on the IEEE 802.11ad protocol at mmWave frequency to validate our analytical results. The ns-3 results show that proposed multi-AP deployments produce significantly higher aggregate performance as compared to other common AP placements in indoor scenarios.

The position of a wireless access point (AP) in wireless networks has been found to have a considerable impact on the overall network performance in indoor scenarios. Recent work has investigated the impact of the AP's location on the received signal strength of clients [1], and has shown that a 1:7x throughput improvement can be achieved by simply moving the AP in a 2ft. x 2ft. region. The benefits of small scale AP mobility is chiefly caused by mitigating multipath effects. In fact, the multipath effect has a significant impact on network performance and can be dramatically altered even with mere centimeter level movement of a Tx or a Rx. In this work, we investigate how network throughput performance can be improved if the AP is able to adapt its orientation. We consider two types of orientation changes - that of the AP's base platform (base orientation), and that of its antennas (antenna orientation). We show using experimental analysis that network throughput performance can be improved 1:8x by simply adapting AP's orientation.

We ask the following question in this paper: Can the goals of centralized WiFi scheduling be achieved using purely distributed operations? We present a solution called Look Who's Talking (LWT) that allows for arbitrary schedules to be distributed to nodes in a WiFi network. The nodes in the network then use purely local and distributed operations to achieve the prescribed schedule. The scope of LWT in this paper is restricted to a single collision domain (single or multiple cells), but we discuss how LWT can be extended to multiple collision domains. We use both experimental evaluations (using a WARP-based testbed) and simulation-based analysis (using ns3) to evaluate LWT.