How does LoRa backscatter let battery-free sensors send data over long distances?

The short answer: reflect a radio signal instead of generating one

LoRa backscatter lets battery-free sensors send data over long distances by borrowing energy from a radio wave that is already in the air. A separate device, the RF source, broadcasts a plain carrier. The sensor tag never builds a carrier of its own. It flips its antenna between different impedance states, which changes how much of that carrier it reflects and in what phase, and the pattern of those reflections is shaped so that it looks like a standard LoRa packet to an ordinary LoRa receiver.

The payoff is power. Ordinary radios typically draw somewhere between 10 and 500 milliwatts while transmitting. The University of Washington team that first demonstrated LoRa backscatter reported an integrated circuit that consumes 9.25 microwatts, which they describe as more than a thousand times lower than LoRa radio chipsets, and estimated it would cost less than a dime at scale. At that level a tag can plausibly run from harvested energy alone.

Why making a LoRa chirp by reflection is hard

Classic backscatter, the kind used in RFID tags, sends simple on-off or two-level patterns over a few meters. LoRa is harder. It uses chirp spread spectrum: each symbol is a tone that sweeps linearly across the channel bandwidth, and data is carried in where that sweep starts. A receiver multiplies the incoming signal by a reference chirp and takes an FFT, so each symbol shows up as a peak in one FFT bin. That is what gives LoRa its ability to decode signals far below the noise floor.

To fake this by reflection, the tag must produce a continuously changing frequency, and the number of distinct frequencies needed grows rapidly with the spreading factor. Doing it with banks of oscillators or a very fast clock would cost too much power. The UW design used a hybrid approach: low-power digital logic works out a frequency plan, a DAC drives a voltage-controlled oscillator, and an RF switch network maps the result onto eight impedance states. Using a four-level approximation of the in-phase and quadrature signals suppresses the unwanted mirror image and cancels the third and fifth harmonics that simple square-wave switching would spray into neighboring channels.

Dealing with the loud carrier next to a whisper of a signal

The hardest practical problem is self-interference. In the UW measurements the backscattered packet arrived at the receiver around minus 134 dBm, while the direct carrier from the RF source was around minus 45 dBm, roughly ninety decibels stronger. The fix is to shift the reflected LoRa signal away from the carrier frequency, so the carrier lands outside the LoRa channel and the receiver's tolerance for out-of-band interference rejects it. Regulatory frequency hopping is handled by the source rather than the tag, so the tag can stay simple.

With the source and receiver 475 meters apart, the UW prototype worked at every point between them. With the source placed next to the receiver, packets were decoded up to 2.8 kilometers away. The team also showed coverage across a three-floor house, a 41-room office, a one-acre farm, and even prototypes built into a contact lens and a skin patch.

Variations: ambient LoRa, rectennas and security

A dedicated carrier is not the only option. PLoRa, presented at SIGCOMM 2018, uses ordinary LoRa transmissions already in the air as its excitation. The tag shifts an incoming chirp and splices it into a new valid LoRa chirp on a different channel. The authors reported backscattering to a gateway up to 1.1 kilometers away from a source node 20 centimeters from the tag, but the scheme carries only one bit per LoRa symbol, needs a custom decoder at the gateway, and its packet detection range was about 50 meters.

Newer work folds backscatter into the same antenna that harvests power. A rectenna, a rectifying antenna, turns incoming RF into DC to run the node. Recent designs from a French research group switch one rectenna between harvesting and backscatter modes under digital control, and use backscatter identification as a lightweight security layer. Their reasoning is that a battery-free node can prove who it is without waking its power-hungry transceiver, and they demonstrated replay attacks against LoRaWAN's ABP activation mode to show why such a layer helps. A 2026 paper from the same group in IEEE Transactions on Microwave Theory and Techniques extends the idea to a digitally reconfigurable rectenna that pairs wireless power transfer with LoRa backscatter for secure battery-free sensor nodes.

Where LoRa backscatter falls short today

Backscatter does not remove the need for power, it moves it. Someone still has to run the carrier source, and range depends on where that source sits relative to the tag and the receiver. Data rates are limited to LoRa's already low rates, and the UW authors left large-scale networking of many tags for future work, using simple time-division scheduling in their tests. For sensors that report small readings occasionally, such as soil moisture, temperature, presence or structural strain, those trade-offs can be worth it in exchange for a node with no battery to replace.