Urban Intelligence With Deep Edges

Abstract

With the increased accuracy available from state of the art deep learning models and new embedded devices at the edge of the network capable of running and updating these models there is potential for urban intelligence at the edge of the network. The physical proximity of these edge devices will allow for intelligent reasoning one hop away from data generation. This will allow a range of modern urban reasoning applications that require reduced latency and jitter such as remote surgery, vehicle collision detection and augmented reality. The traffic flow from IoT devices to the cloud will also be reduced as with the increased accuracy from deep learning models only a subset of the data will need to be reported after a first pass analysis. However, the training time of deep learning models can be long, taking weeks on multiple desktop GPUs for large datasets. In this paper we show how transfer learning can be used to update the last layers of pre-trained models at the edge of the network, dramatically reducing the training time and allowing the model to perform new tasks without data ever having to be sent to the cloud. This will also improve the users' privacy, which is a key requirement for urban intelligence applications with the introduction of GDPR. We compare our approach to alternative IoT urban intelligence architectures such as cloud-based architectures and deep learning algorithms trained only on local data. INDEX TERMS Edge computing, transfer learning, deep learning, urban intelligence, IoT, QoS. I. INTRODUCTION Urban intelligence can provide insight into a number of the major problems being faced by our cities, such as air pollution, increased mobile communication network demand, traffic congestion and water floods, etc., through the use of data from IoT sensors and intelligent analytics on the data. This can allow for the efficient running of a smart city, making the most of the resources available. Problems in smart cities such as increased network demand and crowd congestion are set to get increasingly difficult with the UN predicting 60% of people globally will live in cities with at least half a million inhabitants by 2030 [1]. This will exacerbate the current problems in our cities, so it will be critical to mange these problems with intelligent urban reasoning algorithms and a suitable deployment architecture. IoT devices will enable access to a range of devices and information that was not available to urban authorities before. This additional information from air pollution sensors, CCTV cameras, vehicles and so on will allow the development of more advanced urban reasoning applications [2]. The associate editor coordinating the review of this manuscript and approving it for publication was Hao Ji.

A Survey on Deep Transfer Learning to Edge Computing for Mitigating the COVID-19 Pandemic

“…They have used the concept of transfer learning from cloud to intermediate node to edge. In another study [85], Gary White and Siobhan Clarke have proposed a deep transfer learning-based edge computing for urban intelligent systems. They have also used VGG16 pretrained network at edge devices and experimented to classify Dog vs. Cat images.…”

The IoT as a Key in the Sensitive Balance between Development Needs and Sustainable Conservation of Cultural Resources in Italian Heritage Cities

“…On a technical level, efforts to achieve a compromise between complexity and manageability of smart infrastructures can also be found [56]. The combination of cloud computing with strategies to reduce the training times on large Graphics Processing Units (GPUs) by strengthening the potential of network edges is also an interesting point [57].…”

A Survey on Deep Transfer Learning and Edge Computing for Mitigating the COVID-19

“…They have used the concept of transfer learning from cloud to intermediate node to edge. In another study [84], Gary White and Siobhan Clarke have proposed a deep transfer learning-based edge computing for urban intelligent systems. They have also used VGG16 pretrained network at edge devices and experimented to classify Dog vs. Cat images.…”