Above the Cloud

互联和自动车辆应用的设计期望与云的无处不在的通信，并将受到网络覆盖不足的严重影响。做出关键决策所需的数据必须通过云访问，但任何移动网络不可避免的不可预测性和不可靠性可能导致数据在最好的情况下不相关，在最坏的情况下可能会有潜在的危险，并且数据无法及时往返到云来处理需要大量算法处理的大量信息，这些处理对于能够感知和导航车辆周围环境的决策需求至关重要。虽然更复杂的射频(RF)调制可以提高数据速率，软件多路复用可以增加同时连接的数量，但基本的用户-设备-基站空中接口与移动电话网络的早期保持不变。由于在射频频谱的低端缺乏带宽，因此需要更高的频率来传播这些更快的数据速率载波。天线分集技术，如MIMO，可以改善射频路径损耗并减少干扰，但用户保持在基站/接入点的范围内的基本需求仍然是最重要的。不幸的是，随着传感器和软件算法的增加，自动驾驶系统(ADS)的复杂性正以前所未有的速度增长。不幸的是，在较高频率下的整体路径衰减要大得多，因此需要大幅增加基站的数量以提供覆盖。与需要管理多个跨域和处理能力的集中式网关的传统汽车嵌入式处理相比，所需的计算处理相当于服务器性能。这种整合需要软件架构的巨大变化，以便在不同的域之间以非常低的延迟安全地处理和分发数据。目前的估计预测，全自动驾驶汽车将需要10亿行代码来进行处理，以提供情景感知以取代和增强人类驾驶员的行为。目前对通信的要求阻碍了我们达到和生产安全的4级和5级自动驾驶汽车。上述云项目将开发一个互联汽车软件定义的车辆架构，能够处理大量数据流并近实时处理。

Connected and Autonomous Vehicle applications are being designed with the expectation of ubiquitous communication to the cloud and will be severely affected by inadequate network coverage. The data needed to make critical decisions have to be accessed via the cloud, but the inevitable unpredictability and unreliability of any mobile network may render that data irrelevant at best and potentially hazardous at worst and data cannot make a round trip to the cloud in time to process the large amount of information requiring a large amount of algorithmic processing that essential to be executed in real-time for decision making demands capable of sensing and navigating the environment around the vehicle. Transferring such massive amounts of data has proved to be prohibitively expensive.While more complex Radio Frequency (RF) modulation can increase data rates and software multiplexing can increase the number of simultaneous connections, the basic user-device-to-base-station Air-Interface remains unchanged from the earliest days of mobile phone networks. Due to the lack of bandwidth at the lower end of the RF spectrum higher frequencies are needed to propagate these faster data rate carriers. Antenna diversity techniques, such as MIMO, can improve RF path loss and decrease interference but the fundamental need for a user to stay within range of a base station / access point is still paramount. Unfortunately, overall path attenuation at the higher frequency is far greater and so there will be a need to substantially increase the number of base stations in order to provide coverage, due to the cell tower densification issue.The complexity of Autonomous Driving Systems (ADS) is increasing at an unprecedented rate, with increased sensor and software algorithms. The computational processing required is equivalent to server performance than traditional automotive embedded processing requiring centralised gateways managing multiple cross-domain and processing capability. This consolidation requires a dramatic change in software architecture for the data to be processed and distributed securely and safely, at very low latencies, between various domains.Current estimates predicts fully-autonomous vehicles will require one billion lines of code for the processing required to provide the situational awareness to replace and enhance the human driver behaviour. The requirement for communication as currently preventing us from reaching and producing safe Level 4 and Level 5 autonomous vehicles.The Above the Cloud Project will develop a Connected Car Software Defined Vehicle architecture capable of handling the vast data flows and processing it in near real time.