Coded Caching and Interference Management for Future Wireless Networks

将物理世界的数据转化为数字内容和服务，物联网（IOT）将把所有可以连接的事物连接在一起，从而发展成为万物互联。汽车、交通信号灯、停车计时器、无人机、游戏机、电器、传感器等远比人多。虽然以人为中心的网络的“指数”增长必然会在与世界人口成比例的水平上饱和，即，5G网络中的节点数量将超过数万亿。物联网设备的数量已经超过世界人口，并将在2020年初达到地球人口的20倍。这种令人眼花缭乱的增长带来了相当大的挑战和机遇。虽然物体密度的增加会造成巨大的干扰，但在一个小区域内有许多设备可用，可以通过合作通信实现资源共享。在智能设备存储容量快速增长的情况下，协作缓存可以提供高质量的数据传输，而不会对回程容量或移动终端的功率产生不适当的需求。智能设备可以使用设备到设备（D2 D）通信来共享存储的流行内容，而无需通过基础设施进行路由。对象将被分为内容助手（CH）和内容助手（CR），其关系可以被看作是一个二分匹配问题，并使用匹配理论来解决。作为移动的并且具有不同的优先级，CH可能不总是准备好帮助给定的CR。由于CH不可用而导致的内容丢失可以使用纠删码来避免。*在合作缓存的主要限制是：速率，存储和复杂性。我们将找到三者之间的最佳折衷，并针对各种计算约束开发速率-内存权衡。我们还将开发新的强大的缓存方案，使用无比率代码进行有效的内容放置和恢复。* D2 D业务的快速增长将使频谱饱和，从而需要使用非正交多址（NOMA）。我们将为码域NOMA开发新的多址接入、编码和干扰消除方案。*大多数智能设备配备有多个天线，可以用于形成多个波束。这允许使用干扰对准（IA）来增加5G中的D2 D通信的容量。 使用博弈论方法，我们将为D2 D应用开发新的IA方案。在5G中使用超密集网络（UDN）将导致剧烈的小区间干扰，并且将需要复杂的干扰消除技术。为了降低设备复杂性，可以使用云无线电接入网络（C-RAN），其中在连接到云的基带单元（BBU）处执行复杂功能。BBU的内容缓存可以减少回程-前传流量。*我们的研究将包括：用于D2 D通信的协作编码缓存、用于5G网络的协作NOMA、用于D2 D通信的IA以及用于C-RAN的内容缓存。**

Transforming physical world data into digital content and services, Internet of Things (IOT) will connect together all things that can be connected evolving into the Internet of Everything. There are far more cars, traffic lights, parking meters, drones, game consoles, appliances, sensors etc. than people. While the “exponential” growth of human-centric networks is bound to saturate at a level proportional to the world population, i.e., several billions the number of nodes in 5G networks will grow beyond trillions. The number of IOT devices has already exceeded world population and will reach 20 times earth's population in early 2020's. This dizzying growth comes with considerable challenges and opportunities. While the increase in objects' density causes huge interference, the availability of many devices in a small area allows resource sharing through cooperative communications.***Given the rapid growth of storage capacity of smart devices, cooperative caching can provide high quality data transmission without undue demand on the backhaul capacity or mobile device's power. Smart devices can share stored popular contents using Device to Device (D2D) communications without routing through the infra-structure. The objects will be divided into Content Helpers (CH) and Content Requesters (CR) whose relationship can be viewed as a bi-partite matching problem and solved using matching theory. Being mobile and having different priorities, a CH may not be always ready to help a given CR. Content loss due to CH unavailability may be avoided using erasure correcting codes.***In cooperative caching the main constraints are: rate, storage and complexity. We will find the best trade-off between the three and develop rate-memory trade-offs for various computational constraints. We will also develop novel robust caching schemes using rateless codes for efficient content placement and recovery.***The rapid growth of the D2D traffic will saturate the spectrum necessitating the use of non-orthogonal multiple access (NOMA). We will develop novel multiple access, coding and interference cancellation schemes for code domain NOMA.***Most smart devices are equipped with multiple antennas that can be used for forming a multitude of beams. This allows the use of Interference Alignment (IA) for increasing the capacity of D2D communication in 5G. Using a game theoretic approach, we will develop novel IA schemes for D2D applications.***Use of Ultra Dense Networks (UDNs) in 5G will result in drastic inter-cell interference and complex interference cancellation techniques will be needed. To reduce device complexity, Cloud Radio Access Network (C-RAN) can be used where complex functions are performed at Baseband Units (BBUs) connected to the cloud. Content caching at BBUs can reduce the back-haul-front-haul traffic.***Our research will include: Cooperative coded caching for D2D Communications, Cooperative NOMA for 5G Networks, IA for D2D communications, and content caching for C-RAN.**