The swift progression of Internet of Things (IoT) techniques has driven a surge applications in precision agriculture, smart cities, and environment monitoring. These applications, which encompass vast numbers of wireless devices, pose a significant challenge to sustainability, as powering large quantities of devices over extensive areas via power grids or batteries is not scalable. Energy harvesting emerges as a promising solution. As a viable alternative, energy harvesting technology presents a sustainable and scalable solution, utilizing renewable energy sources in the environment such as solar, RF energy, and vibration energy to power IoT devices. Although, this Sustainable IoT (SIoT) approach allows for near-perpetual operation without battery replacements, it necessitates vigilant oversight to ensure efficiency and security. In light of these challenges, my research aims to tackle issues of sustainability, efficiency, and security within the realm of SIoT networks.
SIoT for structure health monitoring
SIoT for structure health monitoring
SIoT for soil monitoring
SIoT for ocean monitoring
Related Publications: IEEE Trans. on Industrial Informatics'23, IEEE Trans. on Network Science and Engineering’22, IEEE Trans. on Mobile Computing’21, IEEE ICC'21, IEEE Trans. on Wireless Communications’20, IEEE Trans. on Communications’20, IEEE Trans. on Sustainable Computing’20, IEEE GLOBECOM'20, Sensors'19, ACM MobiHoc'18
Motivation: For soybeans, cottons and corns, which are the most commonly grown crops in Southern states and Midwest of United States, the majority of water and nutrient uptake occurs in the top 30-60 cm of the soil profile. Insufficient nutrition and poor drained soil will limit the root development of plants, degrading overall plant health and productivity. Therefore, monitoring the soil conditions at 30 cm layer in the long-term and scalable manner is critical in digital agriculture. The goal of this project is to develop battery-free and solar panel-free Subterranean Internet of Things (SIoT) systems to monitor soil parameters. A mobile platform (e.g., UGV, UAV) will be used to charge SIoTs and collect sensing data.
1) Challenges in deep soil charging: Due to the high signal attenuation in soil, the transmit power must be sufficiently high for RF energy to reach SIoTs buried 30 cm deep in the soil. However, the energy supply and payload of the mobile platform are limited. Therefore, an efficient energy emission strategy is required for SIoT charging.
2) Challenges in low-power communication: Active communication modules, which have a substantial power consumption are not practical for use in SIoT. Current RFID systems (i.e., RF backscatter communications) are not suitable for operations at a depth of 30 cm in soil, mainly due to the high signal attenuation within the soil and the considerable depth involved.
Related Publications: IEEE Trans. on Industrial Informatics'23
Nov. 2022 Initial lab testing of RF backscattered communications
Jan. 2023 Field testing of subterranean IoT using backscatter communication
Illustration of Experiment setup
System Architecture:
- Sensors: Sensors are battery-free and cable-free. They harvest energy from the portable reader for sensing and communication.
- Portable reader: Portable reader can be handheld or carried by UGV or UAV. It provides energy to sensors and collects sensing results.
Advantages:
- Easy to use: Sensors are cable free. They can be mixed with concrete or asphalt during building/bridge/road construction.
- No lifetime limitation: Sensors are battery free. They can be implanted permanently into structure without follow-up maintenance (i.e., battery replacement).
This project is collaborative research with Dr. Jun Wang from the Richard A. Rula School of Civil and Environmental Engineering.
We appreciate the support of MMC Materials, Inc, on our field experiments of RF remote charging and backscatter communication in concrete.
Optimal Communication Strategy Design in RF Energy Harvesting Powered IoT
Existing transmission strategies based on unrealistic assumptions are infeasible in practice. Through experimental study, we reveal several practical issues, including the nonlinear battery charging, and dropping supply voltage, in the wireless systems that will challenge the optimal transmission policy design.
Nonlinear battery charging: Due to the nonlinear charging characteristics of supercapacitor, the amount of harvested energy is not a predetermined value as assumed in the conventional model, but a variable depending on the battery’s residual energy. Meanwhile, the residual energy is affected by the data scheduling, which in turn causes a feedback loop from data transmission to harvested energy. Without considering the dependency of energy harvest on data transmission resulting from the nonlinear charging feature of battery, the existing data transmission scheduling strategies that are based on inaccurate estimation of the harvested energy is thus infeasible in reality.
Dropping supply voltage during successive transmissions: Due to the limited energy storage capacity of IoT nodes (e.g., 1 mF capacitor) resulted from low RF harvesting rates, the voltage of the energy storage, which also serves as the IoT’s supply voltage, may experience a significant drop during successive transmissions. Besides communications, other energy consuming operations, such as ADC sampling and computing, will also result in a significant falling of supply voltage. Conventional energy consumption model that assumes constant supply voltage will be inaccurate.
In our research, we put forward a nonlinear model for energy harvesting, devised to precisely predict the amount of energy to be collected for effective transmission scheduling. Moreover, we incorporate the hardware characteristics of IoT devices into our model to accurately estimate their energy consumption.
Related Publications: IEEE Trans. on Wireless Communications’20, IEEE Trans. on Communications’20, IEEE Trans. on Sustainable Computing’20, IEEE GLOBECOM'20, Sensors'19, ACM MobiHoc'18
RF energy harvesting system
Nonlinear energy harvesting model
Depiction of optimal transmission scheduling in new energy harvesting model
Edge Case-Enhanced Active Noise Cancellation for TWS Earbuds
Related Publication: IEEE/ACM Transactions on Audio, Speech, and Language Processing'22, Provisional patent serial number: 63/230,340
In some applications like radio-frequency identification (RFID), a low-power energy harvesting node (EHN) could receive energy from a dedicated RF energy source for wireless communications. How to request energy efficiently is a critical problem. Intuitively, the EHN needs to pay a cost for the transmission of energy requests, thereby producing considerable energy overhead if a node requests over frequently. By contrast, requesting a large amount of energy each time reduces the overhead, but leads to energy inefficiency due to the nonlinear charge characteristic of EHNs. An optimal strategy is necessary for EHNs to request an appropriate amount of energy at the right time. To solve the problem, a path-oriented method is a feasible solution.
In the path-oriented method, an EHN first calculates the least transmission power required for a timely delivery of data collected from biosensors. Afterward, an energy tunnel (yellow region) is formed, as shown in the right figure. The lower bound of the tunnel is determined by the least required energy accumulated with time, and the upper bound is in parallel with but Em above the lower bound, where Em is the maximum energy can be stored in an EHN. The accumulation of harvested energy (red arrow lines in the figure) is subject to the two bounds; otherwise, the harvested energy will either be insufficient for a timely data transmission (below the lower bound) or overflow from the energy storage (over the upper bound).
To find the optimal way for energy requesting, the energy tunnel in the right figure is divided into multiple grids, forming a graph with a set of “vertices” and “edges”. The vertical edge indicates an energy replenishment, which generates an associated charging cost at the energy source. The cost consists of two parts: a constant overhead, and a nonlinear charge cost. The latter is determined by both the residual energy of the EHN, which is the vertical distance to the lower bound, and the amount of energy requested, which is the length of the vertical edge. A horizontal edge means no energy charging and generates no cost at the energy source. Eventually, the optimal energy requesting strategy is converted to find the route between the start point and destination that has the minimum sum-cost along the path. In the right figure, the path is not allowed to move backward or downward since the cumulation of harvested energy increases monotonically with time. Consequently, the dynamic programming methods like Dijkstra’s algorithm can be applied to schedule the optimal energy request for EHNs.
Related Publications: IEEE Internet of Things Journal’18, IEEE INFOCOM’17
An underwater cognitive acoustic network (UCAN), which consists of bottom nodes, autonomous underwater vehicles (AUVs) and surface nodes in the ocean, to detect the presence of interesting targets. When no target is detected, the nodes stay in the sleep mode to save the energy. In this case, each node generates only a few packets periodically to update their routing table or to synchronize time. Therefore, a narrow frequency band for the control channel would be sufficient to guarantee a low collision probability. Other vacant frequencies could be saved for the data transmission to reduce the end-to-end delay. The sensor nodes become busy once a target is present, thereby creating a bursty traffic. In this situation, the nodes need to extend the bandwidth of their control channel, which is physically isolated from the data channel for control message transmissions, to prevent control messages from colliding with each other. Otherwise, the low delivery ratio of control message would become the bottleneck of network performance by affecting the transmission opportunity of data packets.
In order to achieve a good throughput and energy efficiency, an underwater cognitive MAC should be capable of adaptively balancing the bandwidth between the control channel and the data channel based on the real-time traffic of a network. However, most cognitive MAC protocols do not have this capability, which promotes us to design a new one for UCAN.
I propose a dynamic control channel MAC (DCC-MAC) for distributed UCANs. Nodes in DCC-MAC could adjust the bandwidth of their control channel adaptively based on the situation of network traffic. Whenever acoustic nodes detected the congestion of CCC, they could flexibly select proper data channels to extend the bandwidth of their control channel and return excessive frequency bands back when the control channel becomes idle.
Related Publications: IEEE Trans. on Mobile Computing’17, IEEE INFOCOM’16, IEEE Trans. on Emerging Topics in Computing'14, IEEE SECON'14
For the received signal strength based (RSS-based) key generation approach, the communicating parties on the two ends of a reciprocal link can produce a shared key through local RSS measurements. An adversary that is monitoring the communication channel, however, can hardly guess the secret key if it is physically near neither communicating entities. The security is consequently ensured with the spatial diversity of a wireless channel, as shown in the left figure, where Alice and Bob are two communicating parties while Eve is an eavesdropper.
Currently, little is known about the actual performance of existing RSS-based key generation methods in oceans, and no attempt has been made in the literature to evaluate them with sea experiments. We conducted a series of experiments from Nov. 2013 to Apr. 2014 to evaluate the performance of different RSS-based approaches under different sea conditions.
From the experiment results, we observe that:
The transmission time of a probe signal to measure RSS in underwater acoustic networks (UANs) is much longer than that in radio networks and thus results in a low key generation rate.
Due to the long propagation delay and large transmission time, the asymmetry of RSS measurements between two communicating parties is more significant in UANs than in radio networks, which causes a high bit mismatch rate on the shared key.
How to improve the performance of RSS-based key generation approaches in ocean environments was studied carefully in my work.
Related Publications: IEEE Communications Magazine'16
Oceans are complex environments. Different kinds of acoustic users, including sonars, marine mammals, and sensor networks, need to share the limited acoustic channel for navigation, foraging, and communications. How to make underwater acoustic networks (UANs) use the underwater channel environment-friendly with other acoustic users is an interesting problem. Cognitive acoustic (CA) is a promising technique to solve this problem.
In UCANs, users are capable of sensing the surrounding environment, and then dynamically configure their operating frequency, transmission power or other system parameters to avoid the interference with other acoustic users. The cognitive acoustic (CA) technique provides an opportunity for UANs to effectively and friendly use the acoustic channel.
However, there are still many grand challenges on UCANs due to the unique features of underwater systems, such as the long preamble signal of an acoustic modem, a large propagation delay of acoustic signal and unknown signal pattern of marine mammals. How to solve challenges are still open issues.
Related Publications: IEEE Trans. on Mobile Computing’17, IEEE INFOCOM’16, IEEE Trans. on Emerging Topics in Computing'14, IEEE SECON'14
In cooperative communication networks, each node plays a role as a potential relay to help with the packet delivery. The destination combines the signals from both the source and the relay, and thus achieves spatial diversity or multiplexing gain to increase the communication reliability or bandwidth efficiency of the network.
How to select the best single/multiple relays to improve the performance of the network is one of the important problems that need to be concerned with cooperative communication.
In my work, I proposed a new criterion for the best relay selection in UAN. The new criterion takes into account not only the channel quality but also the long propagation delay to improve the network performance in terms of throughput and packet delivery ratio.
I also designed a handshake-based MAC protocol for the new relay selection criterion. In the protocol, a transmission of a source node could switch between a cooperative scheme and a non-cooperative scheme smartly based on the packet size, channel quality and the position of potential relays in each round of communications. Therefore, the new protocol guarantees that the performance of a cooperative communication is higher than or, at least, equal to non-cooperative communications.
Related Publications: IEEE MASS'13
In conventional MAC protocol for UANs, people mainly focus on improving the network performance after considering the features of acoustic channel, like the long propagation delay, high dynamic of multipath response and large packet loss rate. How the features of acoustic modem affect the performance of MAC protocols, however, are still not well investigated. According to the experiment results, several unique features of a real acoustic modem, e.g., AquaSent OFDM and Teledyne Benthos, were observed:
The long preamble sequence before a data transmission
The low transmission rate for acoustic communications
I am interested in how these unique features affect the throughput, end-to-end delay, and energy efficiency of a MAC protocol in oceans.
Related Publications: IEEE Wireless Communications Letters'17, IEEE OCEANS'12
In the Atlantic Ocean experiment, we tested three representative MAC protocols in a multi-hop string network. The test lasted for 5 days from September 6 to September 10, 2012. In the experiment, we deployed 9 nodes with Teledyne Benthos modem, forming a string topology in the target region. The strip area of this experiment was about 120 kilometers off New Jersey shore with an average water depth of 80 meters. The acoustic modems were deployed about 30 meters below the sea surface. The weather was rated from moderate to rough during the experiment with wave height between 1.25 and 2.5 meters. Three features of an acoustic modem and underwater channel are observed in the experiment:
1) High packet loss rate and significant channel asymmetry
2) Communication range uncertainty
3) Delayed data transmission
Related Publications: Computer Communications'15, IFIP Networking Conference'13 (Best Paper Award)