Natural Language Processing (NLP) for ITS
As the world’s urban population rapidly grows, cities all over the world are experiencing severe traffic congestion. Traffic congestion causes widespread socioeconomic and environmental issues for cities. According to a study of the Partnership for New York City, the New York City economy is expected to incur losses of over $20 billion over the next five years because of traffic congestion that leads to unforeseen business expenses, wasted fuel consumption from increased idling, and increased vehicle maintenance costs. In addition to the rapid growth of the number of vehicles on the road, unexpected roadway conditions caused by roadside construction/maintenance, car accidents, asynchronous traffic signals, changes in driver behavior, among other factors also cause the traffic congestion in urban areas. The propagation of traffic disturbances along with a lack of rapid information sharing between drivers can easily lead to widespread delays in an entire city’s traffic network. To address the root problem of urban traffic congestion, many technical solutions have been proposed to provide real-time traffic information to drivers. This information can lead them to reroute their trip to avoid congested traffic links in advance. Applying Natural Language Processing (NLP) in social media for the purpose of leveraging underutilized transportation-related posts for Intelligent Transportation Systems (ITS) applications is an active area in research. In this project, we propose to support and complement traffic reporting systems and navigation assistants by exploiting the abundance of data in social media and use it as an additional source of traffic information. To this end, we develop an automated framework that automatically processes social media data from the web, classifies it, and extracts traffic-related reports to convert them into navigation alerts. One situation where this framework has the ability to effectively work is the case when drivers are stuck in heavily-congested roadways. There is a tendency that users could post about the situation in social media, which is meaningful for other drivers. However, this is not the only unique possibility. The framework also uses inputs from specialized agencies and from regular people who are not necessarily driving such as cyclists, pedestrians, or passengers in vehicles, ride-sharing taxis, or buses, etc.
The application can also employ speech-to-text technology so drivers could safely share information in real-time to social media platforms while driving. Another advantage of this framework is that it does not require a specific application to share the information, e.g., Waze, where only the app users could share this type of information together. On the contrary, this solution does not require registration in any specific application, and allows the extraction of traffic information from any social media platform and share to any navigation or traffic-related app. For example, regular social media users can be sharing a publication/tweet about a damaged car on the road in front of their buildings without necessarily having the intention to report incidents to other drivers like in Waze. The framework automatically browses social media platforms, distinguishes traffic-related messages, extracts/understands the incidents information, and converts them into alerts in the navigation apps. A joint text processing framework based on fine-tuning BERT classification models for filtering the traffic related information, and either a Question-answering (QA) model or a name-entity recognition (NER) model for extracting real-time traffic details data from social media are developed. In this two-phase NLP pipeline, we first develop a filter that classifies numerous collected text inputs into different groups. Then, we extract necessary information from these filtered groups in order to automatically understand and characterize the reported traffic event from social media by determining its location, its occurrence time, and its nature, e.g., blocked road, accident, etc. Once the detailed traffic information is obtained, we implement an automated system that converts collected posts from social media into real-time traffic information that are used to update navigation maps and warn drivers about any traffic events.
Distributed IoT Systems in Smart Cities
Internet-of-Things (IoT) technology has become highly auspicious to enhance automation, efficiency, and comfort level for users. Indeed, the number of IoT devices has widely increased. It is expected to exceed $8.4$ billion devices in 2020 and reach $20.4$ billion devices in 2022 engendering a tremendous amount of traffic and data sharing among data providers and consumers. Although IoT technology proved its efficiency in different fields and became essential in many applications such as healthcare, remote monitoring, smart homes, and smart agriculture, it remains under continuous upgrades and does not reach its maturity yet. The IoT environment is indeed still vulnerable and can be controlled by hackers who can use moderate security levels of hardware as well as firmware vulnerabilities to control those devices for espionage and eavesdropping. Often, data leakage takes place during transmission, share, or storage of data, which may entail serious problems for the IoT owners and users. Indeed, usually, IoT devices acquire basic security and authentication levels. Moreover, the connected devices can be used and monitored by hackers and cybercriminals to create sophisticated cyber-attacks which may lead to dangerous and fatal consequences. One of the most effective means to secure IoT devices and services is providing an end-to-end secure IoT architecture and uncircumventable access control for IoT devices. Artificial Intelligence (AI) has been widely used in many industrial domains for its efficiency in upgrading IoT devices with sophisticated smart features. In this context, we propose a novel architecture for IoT devices combining blockchain and AI technologies not only in order to decentralize data storage and protect shared data into the IoT network but also to enhance its performance and efficiency against malware and cyber-attacks. This work investigates the power of machine learning techniques as well as the efficiency of blockchain technology in order to improve privacy, data sharing, and security for IoT devices and smart city infrastructure which are vulnerable and can be used for sinful activities. We advocate the use of the permissioned blockchain to share and store the IoT devices data for its compatibility in IoT distributed architecture, where unlike some suggested approaches, the connected devices do not participate in the mining process and decision-making due to their limited computation power. Then, we apply machine and deep learning algorithms such as Artificial Neural Networks (ANN), XGBoost, decision tree, and naive bayes for malware detection and classification, running on nodes participating in the blockchain network to control and detect suspicious activities. Through implementation and simulations, we evaluate the efficiency of the proposed architecture in detecting and classifying cyber-attacks using practical and real-world datasets where decision tree-based models show better performance compared to other state-of-the-art algorithms.
Framework for Green UAV-Assisted ITS
Over the last decade, there has been an explosion of interest in connected/autonomous vehicle development, incorporating more devices into the Internet-of-things (IoT), and leveraging the ever-growing user data into useful action plans. The new autonomous and connected vehicles have a whole suite of new technologies. These new features, paired with new high-speed connections to the Internet and new data-processing methods, can be utilized to provide more detailed, accurate, and useful real-time information that can be invaluable for ITS end-users. ITS has the capability of leveraging this data for safer and more reliable transportation systems. RSUs stand to serve as the main infrastructure backbone of an ITS – they will likely serve as the main connection gateway for smart vehicles. RSUs face a number of challenges in practice, due to their placement around ground-level. One example is seamless communication issues – RSUs may have channel interruptions from other radio signals, and obstacles. Moreover, RSUs are relatively expensive to deploy, due to high installation, i.e., capital expenditures (CAPEX), as well as relatively high operational and maintenance costs, i.e., operational expenditure (OPEX). While RSUs serve to work as the main piece of ITS communications infrastructure, UAVs stand to fill a role of providing the network flexibility – they may provide better communication links than RSUs, and cover areas that either lack RSU coverage due to location, or is experiencing an unusual amount of demand. UAV considerations in ITS are a new concept, and as a result, there are not many systems that plan or schedule UAVs in ITS applications, and even less that also consider them in concert with RSU planning. ITS planners face several challenges when it comes to figuring out how to build a robust network of infrastructure involving heterogeneous ITS components, i.e. RSUs and UAVs. First, RSUs only have a limited effective range, which is further affected by transmission channel-loss. Second, UAVs also are limited by range due to two primary factors: 1) battery power – UAVs must charge, act, then return to their bay station prior to re-use, leading to them only being available in a temporary capacity, and 2) on-board transmitter range – the short-range transmitter connected to the UAV has a smaller effective communication range, limited by similar factors as the limited range of the RSU transmitters. Third, there is a finite (and likely small) amount of financial resources that can be budgeted for the RSU/UAV station installation and operations. Fourth, both efficient energy consumption and varying traffic flow through certain areas need to be considered in tandem. Our numerical results show that the model is feasible for almost every combination of financial parameters, which directly correspond to the coverage, as a larger budget does correlate with more overall coverage. We also found a diminishing-returns effect in the financial sensitivity analysis, showing that after a certain point, the benefits of increased financial input into the system become negligible. Another interesting finding is the sudden increase of the rate of change increase in financial scenarios with lower marginal operating costs – implying the change in financial parameters causes the heuristic to find a different local optimum