as) United States
`a2) Patent Application Publication 0) Pub. No.: US 2006/0161377 Al
`
` Rakkola et al. (43) Pub. Date: Jul. 20, 2006
`
`
`US 20060161377A1
`
`(54) LOW POWER MOTION DETECTOR
`
`Publication Classification
`
`(76)
`
`Inventors: Juha Rakkola, Espoo (FI); Jukka
`Salminen, Espoo (FI); Kimmo Koli,
`Helsinki (FI); Teemu Salo, Klaukkala
`(FD
`
`Correspondence Address:
`WARE FRESSOLA VAN DER SLUYS &
`ADOLPHSON, LLP
`BRADFORD GREEN, BUILDING 5
`755 MAIN STREET, P O BOX 224
`MONROE,CT 06468 (US)
`
`(21) Appl. No.:
`
`11/027,024
`
`(22)
`
`Filed:
`
`Dec. 30, 2004
`
`(51)
`
`Int. Cl.
`(2006.01)
`GOIP 15/00
`(2006.01)
`GO6F 15/00
`(52) US. Ce ec ceeesecssncesecsssscecesencesenessscecsenceseeneaee 702/141
`
`(57)
`
`ABSTRACT
`
`An energy-efficient acceleration measurement system is
`presented. The system includes an accelerometer, responsive
`to acceleration of the system, for providing an accelerometer
`output signal having a magnitude indicative of at least one
`componentof the acceleration. A motion detectoris respon-
`sive to the accelerometer output signal, and provides a
`processor interrupt signal, but only if the magnitude of
`acceleration reaches a threshold. The processor, responsive
`to the processor interrupt signal, measures the acceleration
`with higher accuracy than the motion detector is capable of,
`but in a way that consumes more powerthan was needed by
`the motion detector.
`
`X-axis Output
`
`Full Accuracy
`
`
`
`Front end
`Digital Signal
`Bus Interface
`i
`Analog
`to
`Processin
`Capacitive
` vexsoupt| mai
`Digital
`With
`(SPII2C,etc)
`
`
`
`Capacitance
`gia’
`ri
`Accelerometer L2%'s Outout
`
`Conversion
`Filtering,
`—_
`to Voltage
`
`Decimation,
`Application
`Z-axis Output
`Conversion
`Calibration
`Processor
`
`
`
`
`
`
`
`SensorInterface and Signal
`
`
`Processing Application-Specific
`
`Integrated Circuit (ASIC)
`
`
`
`
`Low Power
`Motion
`Detector
`
`
`
`
`APPL-1006 / Page 1 of 11
`Apple v. Uniloc
`
`APPL-1006 / Page 1 of 11
`Apple v. Uniloc
`
`

`

`Patent Application Publication Jul. 20,2006 Sheet 1 of 5
`
`US 2006/0161377 Al
`
` ydns9a}uyJOSS8D01gAsandjndjnoINdjNOpewenH J0}99}9q
`JeuBbisJeubisjeuBis
`
`
`SLLSOL
`
`
`
`3NdjNOJoOyesaja00V
`
`JEUDdIS
`
`J@MO0dqMO7
`
`uonoy
`
`aniioedey
`
`JaJawoaja00V"
`
`JOSS8901q
`
`O€L
`
`
`
`
`
`
`
`
`
`L‘Sls
`
`OOL
`
`APPL-1006 / Page 2 of 11
`
`APPL-1006 / Page 2 of 11
`
`

`

`Patent Application Publication Jul. 20,2006 Sheet 2 of 5
`
`
`
`
`
`soe}9}usngjeuBlgjey61gpuajuolndingSKE
`
`
`
`
`
`Aoeinooy
`
`||nN4
`
`
`
`
`uoneojddy‘uUOI}eWUIDEGpUOISIBAUDDndjnosixe-z
`
`JOSS890)quoljeiquedO12
`
`
`
`‘Buvey|i4UOISIAUODoSabe}/0A0}
`
`
`(938‘OZI'IdS)Buissed0l40}BoyeuyUM™
`WIMley61qeoueyoedegndinoSKe-Aompedes
`
`
`
`
`10}99}2q—UoROW”JaMOdMO}
`GLI002
`
`US 2006/0161377 Al
`
`GL2
`
`
`
`jeubigpuesoepayuyJoSuaS
`
`
`
`oyloadg-uoneoiddyBuissao0dd
`
`
`
`(DISW)WNUIDpayes6o}UI
`
`¢Sls
`
`APPL-1006 / Page 3 of 11
`
`APPL-1006 / Page 3 of 11
`
`
`
`
`

`

`Patent Application Publication
`
`Jul. 20, 2006 Sheet 3 of 5
`
`US 2006/0161377 Al
`
`Ole
`
`GLE
`
`Ode
`
`uOHoOy
`
`
` Ayeonewojny Joyoa]9qBuljqeug
`
` BulpiAdid YUeinjesadwia|10}ayesuadwoDAjjeonewoyny sixyoeyJO4eyeq
`Bulbesany sIx¥yOeF
`
`
`
`
`
`
`
`
`pjoysely|Sspeeoxy[eae]souaiajeysnulj|UONeIajao0yaBesanyJ!|eUuBISydnuayu|JOSsad0Ig
`SO€ €‘Sls
`
`
`oO}aaa]BoUaJajayHulyepdy
`
`JO}[eAe7]BouaIBjoyeBuyas
`uoNelajaooy
`
`GCE
`
`00€
`
`APPL-1006 / Page 4 of 11
`
`APPL-1006 / Page 4 of 11
`
`

`

`Sty
`
`Bey-wns
`Ja\siBay;Buluwins
`
`
`
`wun)Buyjjonuogauedulog
`
`jdnueyupue
`Bunjseyypue1eSHO
`
`OOF
`
`00r
`
`puejas
`
`edwoy
`
`Wun
`
`SIXB-X
`
`Burwwns
`
`SIXe-Z
`
`Patent Application Publication
`
`Jul. 20, 2006 Sheet 4 of 5
`
`US 2006/0161377 Al
`
`
`
`$i9}siGey}O1]U0DpueuonesnByuog
`
`
`
`Jayunogjauueyy
`
`pueJasyoOalu
`
`ouedwog
`
`wun
`
`0€r
`
`Buiw
`
`49}!
`
`
`
`
`
`SUIYDeIS}e}S|OuUOD
`
`
`
`Jo}99}8qUONOW\
`
`j0J}U0D
`
`ajepdy)soueiejyeyUNAA
`
`
`
`slalsibayjeAe7]soUualajJay
`
`
`
`Ja}UNODajdwes
`
`vySls
`
`APPL-1006 / Page 5 of 11
`
`APPL-1006 / Page 5 of 11
`
`
`
`

`

`Patent Application Publication Jul. 20,2006 Sheet 5 of 5
`
`US 2006/0161377 Al
`
`OVS
`
`02S
`
`S¢S
`
`
`
`wingojepdn
`
`SOS
`
`OLS
`
`SLS
`
`Spoy|
`
`JouODBuje490/9|esc
`
`
`=|OPI4010838UOHOW
`
`
`101}09]eqUOHOYJe}Uy
`
`JGA97]BOUBJIAjaYaS
`
`HO40}99}0qUOKHOW
`
`
`
`BAIOY}dNwWea}U]
`
`02S
`
`¢‘Old
`
`0eS
`
`oO00S
`
`APPL-1006 / Page 6 of 11
`
`APPL-1006 / Page 6 of 11
`
`
`
`
`
`
`
`
`

`

`US 2006/0161377 Al
`
`Jul. 20, 2006
`
`LOW POWER MOTION DETECTOR
`
`FIELD OF THE INVENTION
`
`[0001] The present invention relates to accelerometers,
`and more particularly to the processing of signals from an
`accelerometer.
`
`BACKGROUND OF THE INVENTION
`
`[0002] Accelerometers are used to convert gravity-in-
`duced or motion-induced acceleration into an electrical
`
`signal that can subsequently be analyzed. Accelerometers
`are used in widely diverse applications including automobile
`air bag and suspension systems, computer hard disc drives,
`detonation systems for bombs and missiles, and machine
`vibration monitors. Accelerometers are also useful in por-
`table devices such as wireless telephones, for example in
`order to cause a device to power up when the.device moves
`sufficiently to indicate that a person has picked up the
`device. Wireless devices such as wireless phones operate
`using small batteries, and therefore it is important for every
`component of a wireless device to consumeaslittle power
`as possible, including not just an accelerometer, but also the
`circuits used to evaluate data from an accelerometer.
`
`[0003] The simplest accelerometeris only able to measure
`one componentofthe acceleration vector, but more complex
`accelerometers are equipped to measure all three compo-
`nents of acceleration, in which case the accelerometer is
`called a 3-axis or triaxial accelerometer. The output signal
`from, an accelerometer may be digital, or the output signal
`may be converted from analog to digital.
`
`It has been well known for many decades that a
`[0004]
`typical accelerometer will include a “proof mass” (some-
`timescalled a “seismic mass”) whichis attachedto a spring,
`and the output signal will then be determined by the position
`of the proofmass. This process of producing an output signal
`is accomplished differently in different accelerometers,
`which may bepotentiometric, or capacitive, or inductive. In
`recent years, other types of accelerometers have been devel-
`oped which, for example, do not require and proof mass at
`all. The minimum size of an accelerometer has been gradu-
`ally reduced over the years, as Micro Electro-Mechanical
`Systems (MEMS) technology has improved. Some MEMS-
`based accelerometers do not even require any movingparts.
`
`Theelectrical output signal from an accelerometer
`[0005]
`must be processed in order to yield conclusions about the
`acceleration experienced by the device that houses the
`accelerometer. According to typical prior art techniques, the
`digital output signal from a triaxial accelerometer will be
`processed with full accuracy by an elaborate and energy-
`demanding process that involves, for example, a combina-
`tion of high-pass and low-pass filtering, decimation, and
`calibration. It would be very usefulif this prior art technique
`could be complemented by a less energy-demanding tech-
`nique for use when less accuracy is required. This would be
`especially useful in portable wireless devices, which usually
`have very limited power capabilities.
`
`SUMMARY OF THE INVENTION
`
`[0006] The present invention describes a way to imple-
`ment a motion detector that can detect acceleration of a
`
`device by very efficiently analyzing the signal from the
`
`device’s triaxial accelerometer. Once this motion is detected
`by an energy-efficient analysis of the accelerometer output,
`then the acceleration can be quantified using the full accu-
`racy of prior art signal processing techniques.
`In other
`words, the present invention roughly divides the analysis of
`accelerometer output into two parts: motion detection and
`motion quantification. The energy-demanding process of
`motion quantification is performed only after an energy-
`efficient process of motion detection has indicated that
`motion (e.g. acceleration) has reached a certain threshold.
`
`[0007] The present invention automatically sets a refer-
`ence level upon start-up, in order to eliminate device orien-
`tation effects. The invention also automatically updates the
`reference level, in order to reduce drift problems. Addition-
`ally, the invention involves summing of samples prior to
`comparison, in order to eliminate higher frequency signal
`impurities.
`
`[0008] Accordingly, there is no main processor activity
`beyond whatis absolutely necessary, and data is only sent to
`the main processor whena previously established movement
`detection criterion has been met. A primary purpose ofthis
`invention is to relieve the main processor of repetitious
`tasks, and to locally handle sensor inputs and transducer
`outputs in a power-efficient way, while interrupting the main
`processor only when necessary.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`[0009] FIG. 1 shows a system including a low power
`motion detector.
`
`[0010] FIG. 2 shows another embodiment of the system
`including a low power motion detector that analyzes all
`three components of acceleration.
`
`[0011] FIG. 3 shows a method according to an embodi-
`ment of the invention.
`
`[0012] FIG. 4 showsa low power motion detector accord-
`ing to an embodimentof the invention.
`
`[0013] FIG. 5 shows a further methodical aspect of the
`present invention.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`[0014] The motion detector of the present invention uses
`acceleration data that has been generated by an accelerom-
`eter accompanied by measurementelectronics. According to
`an embodiment of the present system, full accuracy accel-
`eration measurement is obtained using a 12-bit analog to
`digital converter (ADC), together with digital signal pro-
`cessing which includes low-pass filtering, several decima-
`tion stages, and a calibration algorithm. ADC with 1 mg/1
`LSB is used for a +2 g dynamic range, and if the dynamic
`range changes then the absolute resolution changes accord-
`ingly. The system also includes a motion detector for the
`initial detection of significant motion, utilizing the lowest
`possible power consumption. The full accuracy modewill
`only be used if the motion detector mode indicates that a
`threshold of motion has been reached.
`
`In the motion detector mode, higher level process-
`[0015]
`ing functions (e.g. using an application processor) can be
`kept in an idle state until there is significant movement, at
`which timethe further processing is required. The processor
`
`APPL-1006 / Page 7 of 11
`
`APPL-1006 / Page 7 of 11
`
`

`

`US 2006/0161377 Al
`
`Jul. 20, 2006
`
`of the device housing the accelerometer can thus perform
`other tasks, or no tasks at all, until being interrupted by a
`signal generated by the motion detector when acceleration
`exceeds a predefined limit.
`
`[0016] The motion detector of the present invention helps
`to save power in several different ways. For example, the
`system processor is not needed to continuously monitor
`movement and can therefore stay in an idle state. Also,
`analog and digital signal processing accuracy requirements
`can be relaxed during the motion detector mode, which
`saves additional power. The analog and digital signal pro-
`cessing data rate (operating frequency) requirements can
`likewise be relaxed, in order to save more power. There is no
`need for constant interface activity between the accelerom-
`eter and the processor, which yields further power savings.
`
`[0017] Once movement of the overall device is detected,
`the device can be switched to full accuracy mode, and the
`application processor can be woken up for further analysis
`of movement, or to perform specific actions as a response to
`movement.
`
`[0018] The functionality and properties of the described
`motion detector have several
`important aspects. For
`example, incoming acceleration data is summedinto a single
`register per axis. The number of samples summed in this
`way is a programmable setting. When the programmed
`number of samples has been summed, then the output is
`divided by shifting a bit vector to the left, in order to get an
`average value over a selected number of samples. This
`averaging procedure implements a low-pass filtering func-
`tion, and makes the motion detector insensitive to higher
`frequency signal impurities. This averaging process con-
`sumessignificantly less powerthan filter structures requiring
`multipliers.
`
`[0019] Another important aspect of the described motion
`detector’s embodiments is that, when the motion detectoris
`enabled, a reference level is calculated automatically. The
`benefit of this function is that there is consequently no need
`to consideroffsets on different channels whensetting thresh-
`old levels, and threshold levels can also be set independently
`from device orientation and from the vector of gravitational
`force. An averaging procedure is usedfor this reference level
`calculation as well (see previous description of averaging
`process for incoming acceleration data). The reference levels
`are calculated in this way for each of the three axes,
`assuming that a triaxial accelerometer is used.
`
`[0020] A further important aspect of the described motion
`detector’s embodiments is that the reference level can be
`
`updated automatically and periodically, this period being a
`programmable parameter. This procedure implements a
`high-pass filtering function with very low corner frequency
`(comer frequency is the frequency of the half power point
`which is the frequency at which a filter is transmitting
`one-half of its peak transmission). This procedure also
`reduces motion detector sensitivity to offset drift problems,
`such as temperature drift. Complex machines, robots, and
`wireless devices that require accelerometers function in
`many different environments, and those accelerometers are
`therefore required to maintain their precision as the sur-
`rounding atmosphere changes,
`including alterations
`in
`humidity, pressure, and especially temperature. The refer-
`ence level offset can be updated by a register write operation
`from the processor as well. Combined with the incoming
`
`acceleration data averaging procedure described above, this
`automatic updating of the reference levels implements a
`bandpassfiltering function in a very power-eflicient manner.
`The reference levels are set without regard to device orien-
`tation of the direction of gravity, and so setting of these
`reference levels is greatly streamlined, with corresponding
`reduction of power requirements.
`
`[0021] An additional important aspect of the described
`motion detector’s embodiments is the idea of programming
`the threshold levels for each axis independently. These
`threshold levels are for triggering an interrupt of the main
`processor, and these threshold level parameters establish a
`difference between the level of absolute acceleration com-
`
`pared to the reference level, in order to trigger an interrupt.
`The equations for trigger conditions are as follows:
`current_jyrteference/,,
`4,threshold
`jax’
`>ax'
`Jaycutent_gyteference/.,yythreshold,
`current__,-reference/,,
`,threshold
`az
`az
`/>az
`
`[0022] Yet another important aspect of the described
`motion detector’s embodiments is that different combina-
`
`tions using “AND” and “OR” logical operators can be
`programmedfor individual interrupt conditions on different
`axes, for the purpose of generating the interrupt of the main
`processor in a way that may vary depending upon what
`functions the overall device is being asked to perform, or
`depending upon other factors that may be variable. These
`different combinations can be programmed, for example, by
`setting two parameters for the x-axis, two parameters for the
`y-axis, and two parameters for the z-axis. Each axis can be
`enabled/disabled to form the OR-operation, or be required/
`not required to form the AND function. The following
`combinations are some ofthe possibilities:
`
`[0023] 1.x (enable x, disable y and z)
`
`[0024]
`
`2. y (enable y, disable x and z)
`
`[0025]
`
`3. z (enable z, disable x and y)
`
`[0026] 4.x and y (enable and require x and y)
`
`[0027]
`
`5. x and z (enable and require x and z)
`
`[0028]
`
`6. y and z (enable and require y and z)
`
`[0029] 7.x and y and z (enable and require x, y and z)
`
`[0030] 8.x or y (enable x andy, disable z)
`
`[0031] 9.x or z (enable x and z, disable y)
`
`[0032]
`
`10. y or z (enable y andz, disable x)
`
`[0033]
`
`11. x or y or z (enable x and y and z)
`
`Thus, for example, combination #1 means that only the
`component of acceleration on the x-axis causes an
`interrupt of the main processor, whereas the other two
`components are not factors in this interrupt decision.
`Combination #4 meansthat both the x-componentand
`y-component must reach a necessary threshold to cause
`an interrupt, regardless of the z-component of accel-
`eration. Combination #9 means that either the x-com-
`
`ponent or z-component can cause an interrupt, and the
`y-componentis not relevant.
`
`[0034] The eleven possible combinations listed above are
`not the only possibilities. For instance, sums of squares of
`the conditions could be used. This would allow the system,
`
`APPL-1006 / Page 8 of 11
`
`APPL-1006 / Page 8 of 11
`
`

`

`US 2006/0161377 Al
`
`Jul. 20, 2006
`
`and in particular the interrupt conditions, to function in a
`mannerthat is covariant with respect to at least one axial
`coordinate rotation.
`
`[0035] As already indicated, the motion detector can trig-
`ger an interrupt signal that is used as a level-sensitive or
`edge-sensitive interrupt. This interrupt
`is set when the
`defined rule for interrupt is met(1.e. thresholds are exceeded
`on selected axis/axes). The interrupt can be cleared by
`writing to an interrupt acknowledgeregister. Status regard-
`ing which axes’ acceleration threshold was exceeded can be
`read from the register interface.
`
`[0036] Furthermore, an important aspect of the described
`motion detector’s embodiments is that, when the device
`modeis set to motion detector modeinstead of full accuracy
`mode, the resolution and data rate of the analog front end,
`ADCconverter, and digital processing functions are reduced
`for example from 12 bits to 8 bits. Considerable saving of
`power can be achieved this way, and there is no need for
`accuracy better than 8 bits when detecting the start of
`movementor detecting that there is movement. For analysis
`of movementitself, the device can then be switched auto-
`matically into full performance mode (i.e. full accuracy
`mode).
`
`[0037] Moreover, an important aspect of the described
`motion detector’s embodiments is that the clock signal is
`gated off when the motion detector is in an “off” or “idle”
`state. The clock is also gated off when there is no new data
`to be processed by motion detector. This results in effective
`clock rate of just couple of kHz which enables extremely
`low poweroperation. The motion detector may be in an “off”
`or “idle” state when the device is in full accuracy mode, or
`when the accelerometer is not providing an significant
`output signal, or in other similar circumstances.
`
`[0038] The motion detector of the present invention can be
`used, for instance, to implement a simple step counter. For
`each step, there would be acceleration exceeding the thresh-
`old, if the threshold is set correctly. This acceleration event
`would trigger an interrupt for a processor to update the step
`counter value, possibly in the graphical user interface. In
`another case, hardware connected to the motion detector
`could calculate these steps by itself, without interrupting the
`processor, and the processor reads the step count when
`needed.
`
`[0039] Referring now to the figures, FIG. 1 shows a
`system 100 including a capacative accelerometer 105 which
`produces an accelerometer output signal 110. The low power
`motion detector 115 receives and analyzes the accelerometer
`output signal 110, and if the motion detector determinesthat
`significant acceleration is or may be present, then the motion
`detector sends a processorinterrupt signal 120 to a processor
`125 which is either in an idle state or is performing other
`tasks. If the processor 125 agrees that significant accelera-
`tion is or may bepresent, basedat least upon analyzing with
`greater accuracy the same data that was analyzed by the low
`power motion detector 115, then the processor 125 sends an
`output query signal 130 to the accelerometer 105 in order to
`seek further output from the accelerometer, and the accel-
`erometer then providesthat further output to the processor in
`a queried output signal 135. The processor is then able to
`more fully and accurately analyze the accelerometer output
`data and/or determine actions that need to be taken in
`
`response to the accelerometer output data.
`
`[0040] However,it should be noted that even ifa threshold
`of significant acceleration is reached (or may possibly have
`been reached),
`then it would sometimes be desirable to
`refrain from a more specific analysis by the processor 125.
`An example would be an application for observing the
`degree of activity of a user by simply counting the number
`of times (or rate at which) a threshold (e.g., 100 gm) is
`exceeded, as measured using an accelerometer 105 located
`at a user’s wrist, in combination with the low power motion
`detector 115.
`
`[0041] Turning to FIG.2, this shows a system 200 accord-
`ing to a further embodiment of the present invention, with
`some more detail than in FIG. 1. The capacitive acceler-
`ometer 205 provides output 210 that includes output for each
`coordinate axis. This output is provided to a sensor interface
`and signal processing ASIC 215. This integrated circuit 215
`includes a front end 220 with capability to convert capaci-
`tance to voltage (C-to-V), and this voltage is provided to an
`ADC 225. Note that other types of accelerometers could be
`used instead of a capacitive one, and someofthe alternatives
`are potentiometric and inductive accelerometers. The result-
`ing digital signal 230 is fed to the low power motion detector
`115, which analyzes that data, and provides an interrupt
`signal 240 to both an application processor 245 and a full
`accuracy digital signal processor 250. The processor 250
`then receives further digitized accelerometer output 255 for
`filtering, decimation, and calibration in order to more fully
`analyze the accelerometer output. The processor 250 is then
`able to provide instructions or the like to the application
`processor 245 via a bus interface 260, so that the application
`processor can take appropriate action in response to the
`detected and quantified acceleration.
`
`[0042] FIG.3 showsa simplified method 300 according to
`an embodiment of the present
`invention. A low power
`motion detector located within a device is enabled 305, for
`example whenthe device is recharged, or when nounit in the
`device is detecting any acceleration. Then a reference level
`is set 310 for each of the three coordinate axes, which
`correspond to axes of the device rather than to fixed axes of
`the environment. Setting a reference level may involve, for
`example, adjusting acceleration measurements to offset
`errors caused by things like temperature, air pressure,
`humidity, and other factors that can be taken into account
`even before acceleration is measured. Then accelerometer
`
`data for each of the three axes is averaged 315, which is a
`simple and power-efficient way of deemphasizing measure-
`ment errors. The reference level(s) are automatically and
`periodically updated 320, to compensate for changing envi-
`ronmental conditions. A processor interrupt signal is pro-
`vided 325 if the average acceleration minus the reference
`level exceeds a threshold, so that the processor can then
`monitor acceleration with the usual full accuracy and/or take
`actions in response to the detection of acceleration.
`
`[0043] FIG. 4 depicts an embodiment of a low power
`motion detector 115 according to an embodiment of the
`present invention. X-axis accelerometer data 405 is fed to an
`X-axis summing register 410, and likewise y-axis data 415
`and z-axis data 420 are fed respectively to a y-axis summing
`register 425 and a z-axis summing register 430. Each of
`these three registers also receives input from a sample
`counter 435 and a channel counter 440, which keep count of
`the various sums. The summing registers provide the
`summed averages to respective offset and compare units
`
`APPL-1006 / Page 9 of 11
`
`APPL-1006 / Page 9 of 11
`
`

`

`US 2006/0161377 Al
`
`Jul. 20, 2006
`
`445. The offset and compare units 445 apply offsets provided
`by at least one reference level register 450, and compare that
`result to at least one threshold provided to the offset and
`compare units 445 by at least one configuration and control
`register 455. The reference level register 450 is configured
`to accept updates from a machine 460 which controls the
`reference level updating process. The offset and compare
`units 445 provide the results of their offsetting and com-
`parison to a masking and combining unit 465 that will apply
`a combination such as combinations 1-11 listed above. If the
`result of that combination is detection of a significant
`acceleration, then an interrupt unit 470 is responsible for
`alerting components outside the motion detector 115.
`
`[0044] FIG. 5 shows a method 500 according to an
`embodimentof the present invention. Initially, the motion
`detector is off 505. Upon turning on, the motion detector
`enters motion detector mode 510, and a referencelevelis set
`515 so that data from an accelerometer is appropriately
`offset to compensate for variable environmental conditions
`in which the accelerometer operates. If no significant data is
`forthcoming from the accelerometer (e.g. because the accel-
`erometer data is not changing significantly over time), then
`the motion detector shifts to an idle mode, and may even-
`tually revert to the “off’ mode 505. However, if the motion
`detector does receive significant data from the accelerom-
`eter, then the motion detector updates 520 a sum ofaccel-
`eration data, compares 525 that offset sum to a threshold,
`and activates 540 an interrupt if the result of the comparison
`525is positive (i.e. if a threshold is exceeded). The interrupt
`sends 545 the motion detector back to an idle state, because
`another unit (e.g. a full-accuracy signal processor) will
`become responsible for analyzing the accelerometer data,
`instead of the motion detector performing that analysis. The
`reference level is reset 515 not just when the motion detector
`is started, but also automatically and periodically after a
`specified time has passed,orthis resetting 515 can be forced
`with a register write. A clock gating control 540 controls the
`idling or shutting off of certain areas of the acceleration
`measurement system during periods of acceptable inactivity,
`such as when the motion detector is idled or shut off after
`motion has been detected and an interrupt signal has been
`sent to a processor.
`
`It is to be understoodthatall of the present figures,
`[0045]
`and the accompanying narrative discussions of best mode
`embodiments, do not purport to be completely rigorous
`treatments of the method, system, and apparatus under
`consideration. A person skilled in the art will understand that
`the steps and signals of the present application represent
`general cause-and-effect relationships that do not exclude
`intermediate interactions of various types, and will further
`understand that the various steps andstructures described in
`this application can be implementedby a variety of different
`combinations of hardware and software which need not be
`further detailed herein.
`
`1. An energy-efficient acceleration measurement system,
`comprising:
`
`an accelerometer, responsive to acceleration of the sys-
`tem, configured to provide an accelerometer output
`signal having a magnitude indicative of at least one
`componentof the acceleration;
`
`a motion detector, responsive to the accelerometer output
`signal, configured to provide a processor interrupt
`signal if the magnitude reaches a threshold; and
`
`a processor, responsive to the processor interrupt signal,
`configured to determine acceleration with higher accu-
`racy than the motion detector is capable of
`
`wherein the interrupt signal is for interrupting a power-
`saving state of the processor and for beginning a
`power-saving state of the motion detector, and
`
`wherein the processor is responsive to the accelerometer
`output signal only when the processor interrupt signal
`has interrupted the processor.
`2. The acceleration measurement system of claim 1,
`
`wherein the interrupt signal marks a change from a
`motion detector modeto a full accuracy mode,
`
`wherein, during the full accuracy mode, the processor
`utilizes digital signal processing that includes low-pass
`filtering, a plurality of decimation stages, and a cali-
`bration algorithm, and
`
`wherein the motion detector mode consumesless energy
`than the full accuracy mode.
`3. The acceleration measurement system of claim 1,
`
`wherein the system further comprises a converter for
`converting the accelerometer output signal from analog
`to digital.
`4. The acceleration measurement system of claim 1,
`wherein the processoris also for performing responses to the
`acceleration, but only if the processor has received the
`processor interrupt signal.
`5. The system of claim 1, wherein the system is housed in
`a portable device that also includes a mobile communication
`terminal.
`
`6. A method of analyzing an accelerometer output signal
`using minimal power consumption, comprising:
`
`automatically setting a reference level for each axis, upon
`enabling a motion detector,
`
`averaging acceleration data obtained from an accelerom-
`eter output signal, so as to accomplish low passfiltering
`results without a multiplier,
`
`automatically updating the reference level at periods that
`are sufficiently small to offset temperature drift, and
`
`providing a processorinterrupt signal if the motion detec-
`tor concludes that an absolute difference between an
`
`average acceleration and a corresponding reference
`level exceeds at least one threshold
`
`wherein the interrupt signal is for interrupting a power-
`saving state of the processor and for beginning a
`power-saving state of the motion detector, and
`
`wherein the processor is responsive to the accelerometer
`output signal only when the processor interrupt signal
`has interrupted the processor.
`7. The method of claim 6, further comprising the subse-
`quent steps of the processor performing digital signal pro-
`cessing with low-pass filtering, a plurality of decimation
`stages, and a calibration algorithm.
`
`APPL-1006 / Page 10 of 11
`
`APPL-1006 / Page 10 of 11
`
`

`

`US 2006/0161377 Al
`
`Jul. 20, 2006
`
`8. The method of claim 6,
`
`wherein the at least one threshold is programmed for each
`axis independently, and
`
`wherein exceeding at least one programmed combination
`of the at least one threshold is necessary for generating
`the interrupt signal.
`9. The method of claim 7, wherein the subsequent steps
`are applied to further accelerometer output signals only if
`the processor confirms that the absolute difference between
`the average acceleration and the corresponding reference
`level exceeds the at least one threshold.
`10. Acomputer readable medium encoded with a software
`data structure sufficient for performing the method of claim
`6.
`
`11. A motion detector for analyzing an accelerometer
`output signal with minimal power consumption, comprising:
`
`at least one reference level register, for automatically
`setting and periodically updating a reference level for
`each axis,
`
`a summing register for each coordinate axis, for averaging
`acceleration data obtained from the accelerometer to
`
`yield an average acceleration,
`
`an offsetting and compare unit, for offsetting the average
`acceleration using the reference level, for providing an
`offset acceleration, and for comparing the offset accel-
`eration to at least one threshold, and
`
`an interrupt unit, for providing a processorinterrupt signal
`if the at least one threshold is reached
`
`wherein the interrupt signal is for interrupting a power-
`saving state of the processor and for beginning a
`power-saving state of the motion detector.
`12. The motion detector of claim 11,
`
`wherein the summingregister is used instead of a multi-
`plier,
`
`wherein the reference level is set and updated indepen-
`dently of device orientation and gravitational direction,
`and
`
`wherein the updating of the reference level is performed
`at intervals sufficient to offset temperature drift.
`13. An
`energy-efficient
`acceleration measurement
`method, comprising:
`
`providing an accelerometer output signal, having a mag-
`nitude indicative of at
`least one component of the
`acceleration, to a motion detector, in response to accel-
`eration of the system;
`
`providing a processor interrupt signal if the magnitude
`reaches a threshold, in response to the accelerometer
`output signal; and
`
`determining acceleration with higher accuracy than the
`motion detector is capable of, in response to the pro-
`cessor interrupt signal,
`
`wherein the interrupt signal is for interrupting a power-
`saving state of the processor and for beginning a
`p