Ongoing and past research is described here. See also the Publications page for further info!
Next-generation aerial drones. The upcoming 5G network architecture may lay the foundation for aerial drone applications beyond current physical and technical limits while opening hitherto unforeseen challenges. For example, the ability to rely on a network substrate with predictable latency and reliability may enable Beyond Visual Line of Sight (BVLoS) operation. The sheer number of aerial drones operating on such a network architecture will need ways to orchestrate a shared aerial space while meeting stated performance objectives in terms of flight time and dependability. Meanwhile, technology evolves and new done designs beyond the regular quadrotors become possible. We are researching new drone designs, algorithms, and system approaches across a range of diverse real-world applications, including surveillance, high-definition video streaming, and pick-and-delivery. Part of our work here is in partnership with Vodafone and relies on their 5G experimental deployment in the Milano metropolitan area. [GETMOBILE23, TMC23b, ICIP23, MOBICOM23, TMC23a, PervasiveHealth22, SENSYS22a, SENSYS22b, GETMOBILE20, MOBISYS19]
Dynamic software updates. As software becomes increasingly crucial to power the operation of diverse systems and profoundly impact everybody’s life, its timely and efficient maintenance becomes crucial. Dynamic software updates are performed without interrupting the software execution. They are useful for applications that require frequent updates but must operate continuously, with no perceivable downtime. In mobile robotics, for example, dynamic software updates are useful for resolving software defects that may endanger the robot or its surroundings and to extend the robot’s capabilities by adding new features. We are investigating models, abstractions, and system techniques enabling robust dynamics software updates, ranging from timing analysis to efficient run-time binary replacement. Work in collaboration with Wang Yi (Uppsala University, Sweden). [TECS23, RTCSA23 (Best Paper Award), EWSN22]
Energy-driven computing. As the Internet of Things becomes increasingly pervasive, device footprints shrink and their ability to store energy, either through traditional batteries or through energy harvesting, correspondingly diminishes. Managing energy becomes key to achieving efficient and long-term operation. Although energy has been a fundamental performance metric for several decades on a range of different platforms, energy-driven computers are designed to treat energy availability as a first-class citizen. Energy availability and its fluctuations become a program input like any other, enabling systems to gracefully adapt to the energy dynamics. They may sleep through periods of no energy, endure periods of scarce energy, capitalize on periods of ample energy, or throttle their operation to tame energy oscillations. [MOBIWAC23, IPSN20, IPSN22a, IPSN22b]
Intermittent computing. We are developing techniques to support applications that may be unpredictably interrupted because of energy shortages and must later resume as soon as energy is newly available. Examples are smart buildings and wearable devices, whose energy provisioning may be assisted through ambient energy harvesting and wireless energy transfer. Because energy availability will be erratic, shutdowns and reboots will frequently happen. To ameliorate this, we are seeking answers to three questions. First is how to enable checkpointing of the program state on stable storage with minimal latency and energy consumption. Second is how to determine when and how to intertwine checkpointing with the main application’s processing. Third is what support to offer to developers to manage the possibility that applications be interrupted for a non-negligible amount of time. The work is partly funded through the Google Faculty Award we received in 2015. Work in collaboration with Hamad Alizai and Junaid Siddiqui (LUMS, Pakistan). [EWSN23, ACTA22, SENSYS21, EWSN21, SENSYS20, ENSSYS20a (Best Paper Award and Best Pitch Video Award), ENSSYS20b, TECS20a, TECS20b, EWSN20, SENSYS19, LCTES19a, LCTES19b, LCTES19c, IPSN17, TOSN16, EWSN16]
Mobile drone computing. Unmanned aerial vehicles, ground robots, and aquatic rovers are revolutionizing mobile sensing applications. Compared to mobile phones or connected cars that can only opportunistically sense, these platforms offer direct control over where to sample the environment: the application can explicitly instruct them on where to move. They can thus implement sensing functionality that was previously unimaginable, such as collecting high-resolution imagery of civil infrastructures where satellite views cannot reach or inspecting the sea floor to gain fine-grained environmental data. We are researching how to build the software that powers these systems and how to enable their efficient and dependable execution. Work in collaboration with Kamin Whitehouse (University of Virginia, US). [SENSYS22a, SENSYS22b, GETMOBILE20, MOBISYS19, CACM18, GETMOBILE17, MOBISYS16 (Best Paper Award and ACM SigMobile Research Highlight), DRONET16 (Best Paper Award), GETMOBILE16, SENSYS14]
The wireless bus abstraction. Cyberphysical systems (CPS) are employed at the core of many safety-critical sophisticated applications in a range of domains, from manufacturing to healthcare. Because of this, they need to operate dependably, efficiently, and in real time. Low-power wireless communications bring unquestionable advantages as underlying networking substrate for CPS, yet current low-power wireless protocols only focus on a few of the performance goals and fundamental qualities useful to the design and operation of CPS. Together with Marco Zimmerling (TU Dresden, Germany) and Prof. Lothar Thiele (ETH Zurich, Switzerland), we are developing the wireless bus: a networking abstraction and underlying implementation able to serve the needs of CPS applications with respect to predictable behaviors, adaptiveness against changing application requirements and network dynamics, as well as efficient run-time operation. [CSUR20, TCPS17, SRDS13, MASCOTS13, SENSYS12, IQ2S12]
Directional transmissions in low-power wireless. Smart antennas may greatly improve the performance of low-power wireless communications; for example, by reducing channel contention as the antenna steers the radiated energy only toward the intended receivers, and by extending the communication range at little additional energy cost. Their potential, however, is largely untapped as existing low-power wireless protocols are built on the old adage of omnidirectional transmissions. We are designing new networking solutions that, by modifying existing designs or by applying a clean-slate approach, can fully harness the benefits the of directional transmissions and, at the same time, we are investigating the limitations of this technology in a low-power multi-hop setting. Work in collaboration with Ambuj Varshney (Uppsala University, Sweden), Thiemo Voigt (RI.SE, Sweden), and Gian Pietro Picco (University of Trento, Italy). [MASS16, SENSYS15, SECON13, EWSN13, REALWSN13, REALWSN10]
Software adaptation in cyber-physical systems. CPSes place a computing and communication core in the environment to gather data from, and possibly take actions in the real world. Because of the intimate interactions between the system and the physical world it is immersed in, CPS software is chiefly required to self-adapt against the many and unpredictable environment dynamics. This is difficult to achieve in general, and even more so whenever adaptation decisions are subject to time constraints or developers are to battle against the resource limitations of many existing CPS platforms. We are studying design concepts, language abstractions, and verification approaches to tackle this challenge on a range of different platforms, from tiny sensor nodes to autonomous robot drones. [TAAS18, REACTION16, CORCS14, DCOSS14]
Dependability in wireless sensor networks. Because of cost and minimal invasiveness, wireless sensor networks are created out of resource-constrained devices that tend to be brittle and fragile. The environment bears great influence on their functioning, and yet its dynamics are partly not even understood and in general difficult to predict. This makes it challenging to achieve dependable system operation. We are investigating how to improve the dependable behavior of wireless sensor networks without sacrificing other key performance objectives, such as energy consumption. This requires studying the problem from different angles, including finding ways to understand the fundamental limitations, adapting existing dependable computation models, as well as increasing the system’s predictability and resilience to the environment dynamics. Work in collaboration with Arshad Jhumka (University of Warwick, UK)and Thiemo Voigt (RI.SE, Sweden). [TOSN16, EWSN16, SRDS13, EWSN10, SRDS09]