Skip to content
New
Latest News || 27 October 2026: Dr Soumava Mukherjee's invited talk at Queen's University Belfast joined device-level hardware innovation with channel-level control. || 22 October 2026: I received a Certificate of Appreciation for serving as a reviewer for IEEE IMAS 2026. || 15 October 2026: I presented a poster on beam-domain ML fingerprinting for RIS-aided mmWave localisation at the Centre for Wireless Innovation's annual symposium. || 15 October 2026: CWI's annual symposium brought research and industry perspectives on 6G, ISAC, optical wireless, AI, and future networks to Queen's University Belfast. || 14 October 2026: Erik Nordell of Ericsson visited our Keysight Lab for a live RIS-assisted video streaming demonstration.
Tarek Hassan
Knowledge Hubintegrated sensing and communication6. Localization & Tracking Perspective

integrated sensing and communication learning note

6. Localization & Tracking Perspective

How ISAC supports device-based and device-free localization, tracking, Doppler estimation, and predictive beamforming.

At a glance

  • Track, not just locatePosition plus velocity over time lets the network predict where a user or target will be next.
  • Device-free is harderTargets that carry no radio send no pilots, so the network must find them in weak echoes and clutter.
  • Predict, then correctA tracker blends a motion model with each noisy measurement, giving a smoother and earlier estimate.

From Positioning to Tracking

Traditional positioning often estimates a static location. ISAC goes further: it can estimate position, velocity, direction, and trajectory over time.

Tracking is valuable because communication links are dynamic. If the network knows where a user is moving, it can predict the next beam instead of waiting for the link to fail.

References for this section3

Device-Based Localization

Device-based localization estimates the position of a connected user. The user participates in the protocol and may send pilots or feedback.

Measurements include:

  • Time of arrival.
  • Angle of arrival.
  • Angle of departure.
  • Doppler.
  • Channel state information.
References for this section3

Device-Free Localization

Device-free localization tracks objects that do not carry a radio device. Examples include pedestrians, vehicles, drones, and human motion.

ISAC supports this because the radar-like echo can reveal object presence and motion even if the object is not connected to the network.

  • Device-free tracking: the target sends no pilot and does not cooperate.
  • Receiver tasks: detect weak echoes, separate clutter, match observations across time, and update a trajectory.
  • Comparison: a connected phone can transmit pilots, making its localization problem different.
References for this section3

Doppler and Micro-Doppler

Doppler measures velocity along the radio path. Micro-Doppler comes from small moving parts, such as arms, legs, rotating blades, or vibrating machinery.

Micro-Doppler can help distinguish:

  • Human walking vs vehicle motion.
  • Drone rotor motion vs bird motion.
  • Machine vibration patterns.
References for this section3

Predictive Beamforming

If sensing estimates user motion, the network can predict future beam directions. This is useful for high-speed trains, vehicles, UAVs, and mmWave links where beam misalignment causes fast outages.

References for this section3

Tracking Filters

Tracking usually uses temporal filters such as:

  • Kalman filter.
  • Extended Kalman filter.
  • Particle filter.
  • Bayesian multi-target tracking.
  • Learning-based trackers.

The ISAC waveform provides measurements; the tracker converts them into stable trajectories.

The idea behind all of them is predict, then correct. A motion model predicts where the target should be now; the new measurement says where it seems to be; the filter blends the two, trusting whichever is less uncertain. The lab below runs a Kalman filter on a moving target so you can compare raw fixes with the filtered track.

Tracking a moving target: raw fixes vs a Kalman filter

1. Formula and symbols
x^−=Fx^,P−=FPFT+Q\widehat x^-=F\widehat x,\quad P^-=FPF^T+Q
K=P−HT(HP−HT+R)−1K=P^-H^T(HP^-H^T+R)^{-1}
x^+=x^−+K(z−Hx^−)\widehat x^+=\widehat x^-+K(z-H\widehat x^-)
  • x^\widehat x: estimated position/velocity state
  • FF: motion model
  • PP: covariance
  • QQ: process covariance
  • HH: measurement model
  • RR: measurement covariance
  • zz: noisy fix
  • KK: Kalman gain

−/+ mean prediction/correction.

BSfaint: true path · orange: raw fixes · teal: filter
Tracker
2. Work with the current inputs
predict: x^k∣k−1=Fx^k−1,  Pk∣k−1=FPk−1FT+Qcorrect: K=Pk∣k−1HT(HPk∣k−1HT+R)−1,  x^k=x^k∣k−1+K(zk−Hx^k∣k−1)\text{predict: } \hat{\mathbf{x}}_{k|k-1} = \mathbf{F}\hat{\mathbf{x}}_{k-1}, \ \ \mathbf{P}_{k|k-1} = \mathbf{F}\mathbf{P}_{k-1}\mathbf{F}^{T} + \mathbf{Q} \qquad \text{correct: } \mathbf{K} = \mathbf{P}_{k|k-1}\mathbf{H}^{T}\left(\mathbf{H}\mathbf{P}_{k|k-1}\mathbf{H}^{T} + \mathbf{R}\right)^{-1}, \ \ \hat{\mathbf{x}}_k = \hat{\mathbf{x}}_{k|k-1} + \mathbf{K}\left(\mathbf{z}_k - \mathbf{H}\hat{\mathbf{x}}_{k|k-1}\right)
3. Result
Raw fix error (RMS)1.71 m
Filtered error (RMS)1.76 m1.0× smaller
Beam aimed atpredicted position
A target (faint line) moves across a 80 m × 37.5 m area. Every 0.1 s the base station measures its range (±0.3 m) and angle (adjustable noise) and converts them to a position fix (orange dots). A constant-velocity Kalman filter (teal) predicts each next position from the estimated speed and blends it with the new fix. The beam is aimed half a second ahead: at the filter's prediction, or, with the filter off, at the latest raw fix. Errors are averaged over the whole run.
References for this section3

Main Challenges

  • Clutter from static reflectors.
  • Multi-target association.
  • Occlusion and blockage.
  • Synchronization.
  • Privacy.
  • Real-time processing.
  • Joint optimization with data communication.
References for this section3

Takeaway

ISAC localization is powerful because the network can both communicate with users and sense non-connected targets. The most important advantage is not just knowing where something is, but predicting where it will be next.

References for this section3

Check your understanding

0 of 4 answered

  1. What makes device-free localization harder than device-based?

  2. What does micro-Doppler help with?

  3. In the tracking lab, why is the filtered track smoother than the raw fixes?

  4. How does tracking improve communication?

Complete references and further reading