I-Corps: Efficient Terrain-Based Measurement Gathering for Path Loss Inference

为了应对蜂窝网络需求的多重增长，运营商蜂窝范围的大小已经向下转移，以最大限度地提高网络容量。在这样做的过程中，准确和细粒度的覆盖估计成为空间重用、小区间干扰和小区间平滑切换的关键问题。从历史上看，电信公司已经使用现场实验来估计给定区域的无线传播特性，由于需要人力和设备资源，这是昂贵的。智能手机的普及提供了一个更便宜的选择，通过众包直接从手机用户那里记录现场网络观察。然而，众包通常提供不完整的测量覆盖，以准确推断路径损失。使问题进一步复杂化的是，不同的子区域可能需要不同数量的测量来实现相同的路径损耗表征精度。这个I-Corps团队开发了一个系统，该系统使用地理特征信息（例如，建筑高度，树叶，自由空间）来预测一个地区的测量收集需求，使用统计学习框架，允许网络运营商优化他们的测量收集策略。拟议的工作将降低在特定用户区域推出新一代蜂窝和/或频段时的财务障碍，并增加覆盖的保证。该项目还将增加对利用来自用户的众包信息的信心，而不是依赖昂贵的驱动测试，从而节省人力，并在网络性能方面实现更快的反馈循环。该项目将使团队成员能够发展商业敏锐度，从而提高他们研究的质量和保真度，并培养其他具有商业上可行的智力资本的各级学生。通过参加I-Corps课程，该团队打算清楚地确定当前运营商花费的确切金额，以及客户愿意为拟议服务支付的金额。

To address multi-fold increases in cellular network demand, carrier cell range sizes have shifted downwards to maximize network capacity. In doing so, accurate and fine-grained estimation of coverage becomes a critical issue for spatial reuse, intercell interference, and smooth handoffs between cells. Historically, in-field experimentation has been used by telecommunication companies to estimate wireless propagation characteristics of a given region, which is costly due to the manpower and equipment resources required. The emerging popularity of smartphones offers a far cheaper alternative, recording in-field network observations directly from the cellular users via crowdsourcing. However, crowdsourcing often provides incomplete measurement coverage to accurately infer path loss. Further complicating the issue, different sub-regions can require differing amounts of measurements to achieve the same path loss characterization accuracy. This I-Corps team has developed a system that uses geographical feature information (e.g., building heights, foliage, free space) to predict measurement collection requirements over a region using a statistical learning framework, allowing network operators to optimize their measurement collection strategy.The proposed work will lower the financial barrier to and increase the assurance of coverage in rolling out new cellular generations and/or frequency bands over a given region of users. The project will also allow increased confidence in leveraging crowd-sourced information from users as opposed to relying on costly drive testing, thereby saving manpower and allowing a quicker feedback loop in terms of network performance. The project will allow the team members to develop commercial acumen that could increase the quality and fidelity of their research as well as train other students at all levels that have intellectual capital that could be commercially viable. By participating in the I-Corps curriculum, this team intends to clearly identify the exact amount current carriers spend, and the amount customers would be willing to pay for the proposed service.