Adaptive Sensors Optimize Indoor Transmissions for Smart City Networks

Through various information and communication technologies, smart cities provide services like education, medical care, safety, transportation, and utilities. Many of the services that smart cities are expected to provide rely on accurate 3D sensing of urban spaces, both indoors and outdoors.

To manage the resources in a smart city, multiple lidar devices form sensor networks to collect data about the position of 3D shapes and objects in real time. The network covers blind spots by aggregating point clouds from multiple lidar sensors that have different viewpoints.

Despite substantial progress in multi-lidar sensor networks for outdoor use, the operation of indoor multi-lidar networks continues to be constrained by fluctuations in available network bandwidth. When bandwidth is limited due to other traffic loads, the transmission size needs to be small while including the most important points. To achieve this, adaptive control is required.

Many of the services expected of smart cities rely on accurate 3D sensing of urban spaces, both indoors and outdoors. In this study, researchers tackled a major limitation of existing indoor multi-lidar sensor networks by implementing adaptive data transmission. The proposed technique will enable more lidar sensors to operate properly in indoor spaces, even with fluctuations in available network bandwidth. Courtesy of jurvetson at Openverse.

To minimize the impact of limited bandwidth and optimize the data volume transmitted by indoor lidar sensors, researchers from Shibaura Institute of Technology and Osaka University developed an adaptive, multi-lidar sensor network.

Their system adaptively controls data volume at the individual sensor level, taking into consideration the state of the network in real time. Each lidar unit is connected to a sensor control device that captures, buffers, and transmits the data to an edge computer. The edge computer constantly monitors and aggregates the data streams from each sensor, keeping track of data latency and jitter. At the same time, it monitors each sensor’s delay status in real time. When the computer detects a delay, it sends a notification to the sensor, instructing the sensor to adjust the amount of data it is transmitting.

To allow changes in data volume, each sensor has a filter that can increase or reduce the size of the point cloud being transmitted, based on which regions of the point cloud are of the least importance in 3D space. The status of each region in the point cloud is predetermined by the administrator.

The system maintains as much of the quality of the aggregated point cloud as possible, even when bandwidth is limited, by allowing the sensor to discard the least important points.

The researchers performed two experimental evaluations of the system in indoor environments under a variety of conditions. One experiment validated the effectiveness of the system, while the other experiment verified its practicality.

“The proposed design was evaluated with various load patterns from 100 to 200 megabytes per second, including dynamic loads that differed in size minute to minute,” researcher Kuon Akiyama said. “We found that our system was able to satisfy delay requirements and receive highly important points even if the network was under dynamic load. This suggests that our approach is effective when an object moves within the monitoring area and when the available network bandwidth varies, regardless of the physical speed of wireless communication.”

Indoor multi-lidar networks rely on wireless communication between sensors, and surges in traffic volume can strain the shared bandwidth. The large number of sensors required to monitor an indoor environment can compound the problem. While four or five sensors are typically enough to cover any blind spots in an outdoor setting, ten or more sensors are usually required to monitor indoor spaces where ceilings and other obstructions restrict the placement of sensors.

By sensing the inside and/or outside of various urban spaces, smart cities can help with routing, logistics, traffic management, and other tasks. Real-time, adaptive data transmission in a multi-lidar sensor network can increase the efficiency of smart city systems, helping to make urban living more comfortable and convenient. Indoor and outdoor multi-lidar sensor networks can be used to create virtual replicas of the real world, known as digital twins.

“A sensor network with a large number of lidar sensors can be leveraged to construct a digital twin that covers a wide area,” Akiyama said. “Such extensive digital twins could be continuously updated in real time to capture the movements of people, vehicles, and mobile robots to optimize autonomous driving systems. These optimizations could improve safety and efficiency of autonomous mobility while reducing costs.”

The research was published in IEEE Sensors Journal (www.doi.org/10.1109/JSEN.2023.3287183).