8.1 Movement Tracking Systems: GNSS/GPS, LPS, IMUs, and Optical Tracking

8.1 Movement Tracking Systems: GNSS/GPS, LPS, IMUs, and Optical Tracking

Quantifying external locomotor and mechanical load in team, court, and field sports requires selecting an Electronic Performance and Tracking System ($\text{EPTS}$) that matches the physical environment, movement velocity, and spatial precision demands of the sport. Within the Needs Analysis and Assessment Selection framework (Domain 3, Task 3B), a sport scientist must understand the underlying physics, sampling frequencies, signal-processing filters, and error profiles of four primary tracking architectures: Global Navigation Satellite Systems ($\text{GNSS}$), Local Positioning Systems ($\text{LPS}$), Inertial Measurement Units ($\text{IMUs}$), and Optical Computer-Vision Tracking.

---

Comparing Athlete Tracking Architectures (EPTS)

1. Global Navigation Satellite Systems (GNSS / GPS, $10\text{–}18\text{ Hz}$)

Wearable satellite tracking units are worn in a tight neoprene or elastic harness positioning the receiver between the upper scapulae over the thoracic spine ($T_1\text{–}T_4$) to minimize upper-limb artifact while maintaining sky visibility. While legacy devices relied exclusively on the U.S. Global Positioning System ($\text{GPS}$) constellation at $1\text{ Hz}$ or $5\text{ Hz}$, contemporary professional sports units are multi-constellation $\text{GNSS}$ receivers sampling at $10\text{ Hz}$ to $18\text{ Hz}$ that simultaneously lock onto multiple satellite networks (GPS, Russian GLONASS, European Galileo, and Chinese BeiDou). Note that some devices marketed at "$15\text{ Hz}$" have produced satellite fixes at a lower rate (for example, $5\text{ Hz}$) and filled the gaps by interpolating with accelerometer data. Always check the raw GNSS fix rate; true $10\text{ Hz}$ units are the usual validated benchmark for field sports.

A foundational technical distinction in $\text{GNSS}$ engineering is how instantaneous velocity and acceleration are derived:

• Positional Differentiation ($v = \Delta d / \Delta t$): Calculating velocity by taking the first derivative of consecutive $X\text{-}Y$ latitude/longitude coordinates (and acceleration via the second derivative) amplifies high-frequency spatial noise by an order of magnitude, rendering acceleration metrics highly unstable.
• Doppler-Shift Velocity: Modern $10\text{–}18\text{ Hz}$ sport $\text{GNSS}$ receivers measure the Doppler shift in the continuous carrier frequency of the arriving satellite radio wave caused by the relative motion of the athlete toward or away from each satellite. Doppler-derived velocity is vastly more accurate and exhibits a much lower Coefficient of Variation ($\text{CV}\%$) for instantaneous speed and acceleration than positional differentiation.

Environmental and Kinematic Limitations: $\text{GNSS}$ radio signals travel approximately $20{,}200\text{ km}$ from medium Earth orbit and require unobstructed line-of-sight sky visibility. Signal validity depends on two diagnostic telemetry metrics that sport scientists must audit before accepting session data:

1. Number of Connected Satellites: Valid locomotor tracking requires a minimum of $8\text{ to }12+$ satellites continuously locked.
2. Horizontal Dilution of Precision ($\text{HDOP}$): Quantifies the geometric dispersion of visible satellites across the sky (lower values are better). When satellites are clustered directly overhead rather than spread evenly across the horizon, triangulation error rises. An $\text{HDOP} < 1.0$ indicates ideal satellite geometry; values between $1.0\text{ and }1.5$ are acceptable; and an $\text{HDOP} > 1.5\text{–}2.0$ indicates poor geometry where positional and velocity data become unreliable.

Near high overhanging stadium roofs or sunken bowls, satellite signals bounce off metal girders and concrete facades before reaching the athlete's receiver—a phenomenon called multipath error. Furthermore, even under open skies, $10\text{ Hz}$ $\text{GNSS}$ exhibits elevated measurement error ($\text{CV} > 10\text{–}15\%$) during rapid $180^\circ$ change-of-direction ($\text{COD}$) cuts, curved sprints, and short, explosive acceleration bursts exceeding $4.0\text{ m}\cdot\text{s}^{-2}$, systematically underestimating peak instantaneous acceleration compared to laser or radar guns.

2. Local Positioning Systems (LPS / Ultra-Wideband UWB, $20\text{–}1{,}000\text{ Hz}$)

To overcome indoor signal blockage and stadium multipath degradation, Local Positioning Systems ($\text{LPS}$) replace orbiting satellites with a fixed array of radio-frequency ($\text{RF}$) anchor antennas mounted around the perimeter catwalks or rafters of an indoor arena or stadium. Modern $\text{LPS}$ installations utilize Ultra-Wideband ($\text{UWB}$) frequencies ($3.1\text{ to }10.6\text{ GHz}$) transmitting ultra-short nanosecond pulses across a wide spectrum ($\ge 500\text{ MHz}$). Unlike narrowband signals, $\text{UWB}$ pulses resist multipath phase cancellation and penetrate human tissue and arena obstacles with minimal attenuation.

Athlete tags communicate with the perimeter basestations at $20\text{ Hz}$ to $1{,}000\text{ Hz}$ (commonly $20\text{ to }100\text{ Hz}$ per player when multiplexing a full squad) using one of two geometric triangulation methods:

• Time Difference of Arrival ($\text{TDOA}$): Synchronized master and slave perimeter anchors measure the exact picosecond arrival-time differences of a signal emitted by the athlete's tag to compute hyperbolic intersection coordinates.
• Two-Way Ranging ($\text{TWR}$): The anchor and athlete tag exchange round-trip timestamps to directly compute absolute distance without requiring sub-nanosecond wired synchronization across all perimeter receivers.

Strengths and Trade-offs: $\text{LPS}$ provides centimeter-level positional accuracy ($< 10\text{–}20\text{ cm}$) and significantly lower error during rapid multidirectional cuts and instantaneous accelerations than $\text{GNSS}$. However, $\text{LPS}$ requires expensive fixed infrastructure, meticulous laser-surveyed calibration of anchor coordinates, and cannot travel with the team to away venues or auxiliary training pitches unless identical hardware is installed.

3. Inertial Measurement Units (IMUs, $100\text{–}1{,}000\text{ Hz}$)

Integrated inside virtually every modern $\text{GNSS}$/$\text{LPS}$ vest pod—or worn as standalone sensors in indoor court sports, combat sports, swimming, and ice hockey—is a micro-electromechanical ($\text{MEMS}$) Inertial Measurement Unit ($\text{IMU}$) sampling between $100\text{ Hz}$ and $1{,}000\text{ Hz}$. A 9-axis $\text{IMU}$ integrates three orthogonal tri-axial sensors:

1. Tri-Axial Accelerometer: Measures linear acceleration and gravitational force (in $g\text{-units}$, where $1\text{ }g = 9.81\text{ m}\cdot\text{s}^{-2}$) along the vertical ($Z$), mediolateral ($X$), and anteroposterior ($Y$) axes.
2. Tri-Axial Gyroscope: Measures angular velocity (in $\text{deg}\cdot\text{s}^{-1}$) around the three axes—quantifying torso roll, pitch, and yaw during turns, tackles, and rotational throws.
3. Tri-Axial Magnetometer: Measures local geomagnetic flux to establish absolute compass heading/orientation (though highly susceptible to ferromagnetic distortion from steel beams or reinforced concrete floors indoors).

Because accelerometers capture high-frequency forces at $100\text{ Hz}$ regardless of satellite visibility, sport scientists use them to compute PlayerLoad (also termed BodyLoad or Dynamic Stress Load depending on manufacturer), which quantifies the instantaneous rate of change of acceleration (three-dimensional jerk) scaled for readability:

$\text{PlayerLoad}_{t} = \frac{\sqrt{(a_{y,t} - a_{y,t-1})^2 + (a_{x,t} - a_{x,t-1})^2 + (a_{z,t} - a_{z,t-1})^2}}{100}$

where $a_{y}$, $a_{x}$, and $a_{z}$ represent anteroposterior, mediolateral, and vertical accelerations at consecutive $100\text{ Hz}$ sample points $t$ and $t-1$. Session PlayerLoad is the sum of these instantaneous values across the session. Crucially, through Kalman filtering and sensor fusion of the accelerometer and gyroscope, Inertial Movement Analysis ($\text{IMA}$) isolates micro-movements that occur with near-zero horizontal displacement—such as rugby scrums, American football line-of-scrimmage collisions, basketball vertical jumps, ice hockey body checks, and directional cutting vectors (classified on a $360^\circ$ clock face into forward accelerations, backward decelerations, and left/right cuts).

4. Optical / Computer Vision Tracking ($25\text{–}60\text{ Hz}$)

Multi-camera Optical Tracking Systems (e.g., Second Spectrum, Hawk-Eye, ChyronHego/TRACAB) use high-definition fixed cameras mounted around the stadium roof operating at $25\text{ Hz}$ to $60\text{ Hz}$ paired with deep-learning computer vision algorithms to track the center of mass (and increasingly 3D skeletal pose landmarks) of every player, official, and the ball simultaneously.

Strengths and Limitations: Optical tracking imposes zero wearable burden on athletes, eliminates device battery/downloading failures, and is the only modality that captures opponent formations and ball trajectories in a unified coordinate system during official matches. However, optical systems suffer from player occlusion and ID-swapping when multiple athletes converge in congested spaces (e.g., basketball paint, soccer corner kicks, rugby rucks), requiring human operators or algorithmic stitching that can smooth out peak instantaneous accelerations. Furthermore, optical cameras cannot measure non-locomotor mechanical impacts, isometric grappling, or vertical ground-contact shock, and data interchangeability error exists when merging match optical data with training $\text{GNSS}$ data (optical tracking often records $5\text{–}12\%$ higher or lower high-speed running distance than $\text{GNSS}$ due to differing smoothing filters, requiring linear regression calibration equations).

---

Absolute vs. Relative (Individualized) Speed & Acceleration Thresholds

The Flaw of Fixed Absolute Velocity Bands

By convention, many sports organizations categorize locomotor volume using fixed absolute speed thresholds—for example, defining High-Speed Running ($\text{HSR}$) as distance covered above $5.5\text{ m}\cdot\text{s}^{-1}$ ($19.8\text{ km}\cdot\text{h}^{-1}$) and Sprinting as distance above $7.0\text{ m}\cdot\text{s}^{-1}$ ($25.2\text{ km}\cdot\text{h}^{-1}$). While absolute bands allow practitioners to compare external work against historical positional match benchmarks, they introduce severe physiological and mechanical misclassification across heterogeneous athletes:

• Overestimation in Fast Athletes: For an elite winger with a Maximal Sprint Speed ($\text{MSS}$) of $10.0\text{ m}\cdot\text{s}^{-1}$ ($36.0\text{ km}\cdot\text{h}^{-1}$), crossing $7.0\text{ m}\cdot\text{s}^{-1}$ represents only $70\%$ of $\text{MSS}$—an aerobic/submaximal striding velocity that imposes minimal eccentric hamstring strain.
• Underestimation in Slower or Youth/Female Cohorts: For a tighthead prop, interior lineman, or youth athlete with an $\text{MSS}$ of $7.6\text{ m}\cdot\text{s}^{-1}$ ($27.4\text{ km}\cdot\text{h}^{-1}$), running at $6.8\text{ m}\cdot\text{s}^{-1}$ represents $89.5\%$ of $\text{MSS}$—a near-maximal neuromuscular effort imposing substantial biceps femoris mechanical strain—yet an absolute $7.0\text{ m}\cdot\text{s}^{-1}$ filter records zero sprint distance and zero sprint efforts.

Individualizing Thresholds via MSS, MAS, and Anaerobic Speed Reserve (ASR)

To reflect true relative physiological and biomechanical strain, sport scientists anchor velocity zones to individual physiological landmarks:

1. Maximal Sprint Speed ($\text{MSS}$) Anchoring: Sprinting is individualized to $\ge 85\%$ or $\ge 90\%\ \text{MSS}$, and High-Speed Running to $65\text{–}85\%\ \text{MSS}$. Every athlete is periodically assessed via $30\text{–}40\text{ m}$ timing gates, radar, or validated peak match $\text{GNSS}$ speed to keep their $\text{MSS}$ anchor current.
2. Maximal Aerobic Speed ($\text{MAS}$) & Anaerobic Speed Reserve ($\text{ASR}$): Derived from an incremental aerobic test or time trial (e.g., $1{,}200\text{ m}$ shuttle test), $\text{MAS}$ represents the lowest running speed at which $\dot{V}\text{O}_{2\max}$ occurs (typically $4.2\text{–}5.2\text{ m}\cdot\text{s}^{-1}$). The difference between an athlete's mechanical ceiling ($\text{MSS}$) and aerobic ceiling ($\text{MAS}$) defines their Anaerobic Speed Reserve:

$\text{ASR} = \text{MSS} - \text{MAS}$

Two athletes with identical $\text{MAS}$ ($4.8\text{ m}\cdot\text{s}^{-1}$) but different $\text{MSS}$ ($8.8\text{ vs. }10.0\text{ m}\cdot\text{s}^{-1}$) possess vastly different $\text{ASR}$ capacities ($4.0\text{ vs. }5.2\text{ m}\cdot\text{s}^{-1}$). Prescribing supramaximal high-intensity interval conditioning as a percentage of $\text{ASR}$ (e.g., $v = \text{MAS} + 0.30 \times \text{ASR}$) equalizes anaerobic glycolytic strain and time-to-exhaustion across both locomotor profiles.

Minimum Effort Dwell Time (Duration Threshold) and Signal Filtering

Whether using absolute or relative thresholds, a discrete "effort" (such as a sprint $> 7.0\text{ m}\cdot\text{s}^{-1}$ or a high-intensity acceleration $> 3.0\text{ m}\cdot\text{s}^{-2}$) is only logged by $\text{EPTS}$ software if the athlete maintains the velocity or acceleration above the threshold for a user-defined minimum effort dwell time (duration threshold)—typically set between $0.5\text{ s}$ and $1.0\text{ s}$.

• Setting the dwell time too short ($< 0.2\text{–}0.3\text{ s}$) causes brief trunk-lean oscillations or Doppler noise spikes to register as false-positive acceleration efforts.
• Conversely, changing the software dwell time setting from $0.5\text{ s}$ to $1.0\text{ s}$ (or updating manufacturer firmware that alters the Exponential Moving Average [$\text{EMA}$] velocity filter window from $0.2\text{ s}$ to $0.8\text{ s}$) can reduce recorded high-acceleration and sprint counts by $30\text{–}55\%$ within the exact same raw session file! Consequently, sport scientists must strictly lock dwell-time durations, minimum inter-effort recovery intervals, and firmware smoothing filters across longitudinal seasons.