IoBaLeT: Sustainable Internet of Battery-Less Things

1) Summary of scientific goals IoBaLeT aims to bring the performance (i.e. throughput, scalability and range) of battery-less IoT networks on-par with its battery-powered counterparts, by enabling active rather than passive communications and computing. An end-to-end networking solution for battery-less IoT will be developed that will achieve this goal through inter-device cooperation and cross-layer energyawareness.

It will consume at most 5% of the available memory, CPU and energy resources. IoBaLeT pursues 4 scientific objectives:

[S1] Accurate energy prediction models to estimate the short-term energy budget of a device, encompassing the interplay between energy storage (e.g., (super)capacitors or hybrid capacitors), harvesting (e.g., voltaic, piezo, electromagnetic) and consumption (e.g., computing, radio, peripherals) processes. We target an average energy consumption and harvesting prediction accuracy of 90% based on detailed pre-generated hardware benchmarks, and 80% for benchmarking-free predictions over a time window of up to 10 minutes.

[S2] Hardware design (i.e., antennas, rectifiers) and multi-antenna transmission techniques for highly efficient cooperative SWIPT, able to transmit 1mW of DC power in a single hop in a 5x5x3m room in both line-of-sight (LoS) and non-LoS conditions. Receiver rectenna efficiency > 40% in the input power range of -10 to 10 dBm. Downlink throughput in line with IoT technologies (Bluetooth, Zigbee) with a power conversion efficiency loss < 5%. This will be extended to a hybrid harvesting solution, combining SWIPT with solar and vibration energy to achieve 10mW harvested power.

[S3] Scalable channel access and routing protocols for multi-hop SWIPT-enabled battery-less IoT networks, able to handle the unpredictable intermittently-powered behaviour of battery-less devices. These protocols should support at least 1000 devices connected to an access point over 1 or more hops and should be able to achieve an end-to-end latency bound of 30 seconds, at a packet delivery ratio of 99.9%, assuming 3 routing hops and 1mW energy harvesting efficiency.

[S4] Energy-aware task scheduler for intermittent devices that intelligently decides which application and network tasks to execute at which time, considering task deadlines, data freshness, expected energy consumption of interconnected tasks and available and expected harvested energy. The task execution failure rate is targeted to be at most 5%. A reduction to 2% is expected when cooperatively scheduling tasks across battery-less and cloud edge devices.