Using Body-Fixed Sensors for Swimming Performance Analysis

We discover how analytics is helping swimmers improve their performance


Like most sports, competitive swimming continues to be influenced by the application of new technology. There is probably no better example of this than the introduction (and subsequent banning) of performance-enhancing, non-textile (i.e., polyurethane) swim suits that saw more than 130 world records broken since their introduction prior to the Beijing Olympics in 2008. While coaches and athletes continue to strive to find new ways to improve performance during competition, new technology in the form of small, body-fixed sensors known as accelerometers are currently available that can monitor performance and stroke technique during training.

Accelerometers are tiny, ubiquitous motion sensors (found in automobile bumpers for airbag activation, in iPhones for screen rotation, etc.) that are about the size of an average coin (see Figure 1) and can be easily worn on the body. This is a key factor in swimming, as anything worn on the body must be small enough not to interfere with the swimmer’s stroke or increase drag. Currently, the swimming biomechanics research group at the University of Regina (Canada) under the direction of Dr. John Barden, uses multiple waterproof-accelerometers manufactured by GeneActiv (Cambridge, UK) to conduct swimming biomechanics research and performance analysis. The accelerometer is a fairly simple device that contains a small mass attached to a spring. As the sensor accelerates, the mass is deflected such that the deflection is converted to an electrical signal. The accelerometer is a great tool for quantifying human movement because it can be placed anywhere on the body to directly measure acceleration in three separate directions. While accelerometers have some limitations (they are not good for tracking position, for example), one of their main advantages is that they can provide large amounts of data for long periods of time. This is particularly useful for monitoring human locomotor movements (e.g., running, walking and swimming) that take place (repeatedly) over extended periods of time. The acceleration data gathered from a swimmer produces a periodic (cyclic) pattern that is unique to a particular stroke and to a particular swimmer’s technique. Our research group has developed specialized software that analyzes the signals from several different accelerometers placed on multiple body sites to produce a range of performance metrics that can be used to optimize stroke technique and performance.

One example of a performance analysis metric that can be obtained from an accelerometer is to use it to conduct a basic temporal analysis for each length of the pool (25 or 50 m), focusing on the inter-stroke interval, swim time and turn time. Figure 2 shows a processed (single axis) accelerometer signal for two lengths of a 50m pool while swimming front crawl. It is immediately apparent that there are numerous spikes or peaks in the signal that occur repeatedly for each length. Each peak corresponds to a single stroke cycle and these can be easily counted to determine the stroke count. More importantly, the time interval between each stroke cycle is also easily determined, thereby providing the instantaneous stroke frequency (or rate) for each individual stroke. These intervals are important because they are influenced by several different factors including speed, breathing and fatigue, and as such provide key information about performance when monitored regularly during training. Consequently, our research group has developed an algorithm that can automatically identify and count strokes, which provides the basis for separating the data into individual lengths but can also provide the turn time (i.e., the time from the last stroke into the wall to the first stroke out of the wall). This metric also provides the coach with important information about the speed and effectiveness of a swimmer’s turn.

Another example of a performance metric that can be obtained from accelerometer data in swimming is the analysis of kick frequency and amplitude. As swim coaches are aware, kicking is important not only for propulsion but also for the correct body position in the water and stroke efficiency (i.e., stroke length or distance per stroke). Figure 3 shows representative accelerometer data for kicking obtained from a swimmer performing one length of front crawl. With two accelerometers placed on the wrists and two at the ankles, the timing between arm and leg cycle frequencies is easily determined. One of the most important applications of the kicking frequency metric is to determine the point at which kicking frequency decreases due to fatigue. Using this method, our research group is currently exploring the degree to which a decrease in kicking frequency is associated with a decrease in stroke length in the last few lengths (or metres) of a race (for e.g., a 200m or 400m freestyle).

The previous examples show that accelerometers can be used to record the cyclic patterns of arm and leg movements for any of the four competitive strokes (butterfly, backstroke, breaststroke or freestyle). However, accelerometers can also provide important information about the acceleration of the body (as opposed to its limbs) when placed close to the body’s centre of mass (i.e., at the base of the lumbar spine). Because accelerometers directly measure acceleration, they are able to detect the constant acceleration due to earth’s gravity. Consequently, they can be used as “tilt sensors”, which occurs when the acceleration due to gravity shifts from one direction (or axis) to another. This is how an accelerometer changes the screen orientation in an iPhone by rotating its position from portrait to landscape mode and vice versa. This application can also be applied to swimming (particularly for front crawl and backstroke), with the sensor being used to calculate the body roll angle about the swimmer’s longitudinal axis (i.e., the axis that runs from head to toe). Body roll is an important performance parameter in swimming front crawl and backstroke, as it’s associated with hip and shoulder rotation, both of which are important for generating power in the pull and push phases of the stroke. The determination of body roll angle can be used to assess bilateral asymmetry, a common condition in most swimmers in which the degree of body roll is greater to one side than the other (this asymmetry typically occurs when a swimmer breathes more to one side than the other). Figure 4 shows a typical asymmetric body roll pattern for one length of front crawl (note the slightly greater peak body roll angles to the left side, which are positive values, than the right side, which are negative values). Our research group is currently investigating the degree to which body roll is affected by speed and breathing preference as well as the degree to which stroke asymmetry (caused by an asymmetric roll pattern) affects performance.

Another benefit of a lumbar-attached accelerometer is that it provides information about the forward acceleration of the body through the water, and can therefore act as a stroke efficiency or propulsion indicator. Current projects include the identification of peak forward body acceleration and the corresponding stroke phase in which it occurs (for all four strokes) as well as the acceleration off the block (and before and after entry into the water) at the beginning of a race. These provide additional performance metrics that can be used by the coach to assess a swimmer’s stroke technique and race-start performance.

In summary, this article has presented several examples to demonstrate how accelerometers can be used to effectively monitor and analyze a swimmer’s stroke performance. They can be used as valuable tools from a research perspective to learn more about the biomechanics of swimming and from an applied sport science perspective to provide valuable performance analytic data to coaches and athletes to improve performance. Although the application of this technology is relatively new to the sport of competitive swimming, it is not taking long to realize its full potential. Demand for this information is increasing, to the point where a new company, Stroke Performance Analytics, was created to satisfy requests from coaches and athletes for stroke performance assessment, both in elite competitive swimming as well as triathlon. For further information about the stroke biomechanics research presented in this article and/or services provided by Stroke Performance Analytics, please contact Dr. John Barden at 


Read next:

Internships: The Current Hot Topic In Analytics