The Dynamic Doppler Spectrum Induced by Nonlinear Sensor Motion: Relativistic Kinematics and 4D Frenet-Serret Spacetime Geometry

The Dynamic Doppler Sp ectrum Induced b y Nonlinear Sensor Motion: Relativistic Kinematics and 4D F renet-Serret Spacetime Geometry Bryce M. Barcla y ∗ and Alex Mahalo v † Scho ol of Mathematic al and Statistic al Scienc es, Arizona State University (Dated: Marc h 27, 2026) F undamen tal to the analysis of nonlinear relativistic motion is the precise c haracterization of the induced dynamic Doppler effects. In this work, we analyze the electromagnetic signals observed by non-inertial receiv ers using tw o framew orks to describ e the relativistic motion. W e first consider observ er paths describ ed by higher-order kinematic 4-vectors: relativistic acceleration and jolt. The dynamic Doppler effects of relativistic acceleration and jolt are exp onential spectral broadening and exp onen tial amplitude growth or decay . W e derive compact expressions for the sp ectrum transfor- mation resulting from relativistic acceleration and jolt. The jolt induces nonlinear skew ed chirps in observed signals. Next we consider observ er paths describ ed by the 4D F renet-Serret frame and the curv ature and torsion of the observ er path. W e obtain descriptions of the amplitude and phase fluctuations of the signal in terms of the geometric parameters of curv ature and torsion. Concise, in terpretable descriptions of non-inertial dynamic Doppler effects provide a useful diagnostic and predictiv e to ol for engineering applications including radar, sensing, and comm unications systems. I. INTR ODUCTION The Doppler effect is a foundational elemen t of man y electromagnetic applications. In tracking radar, changes in signal frequencies are used to estimate the motion of targets [1]. In laser co oling, the Doppler effect is used to decrease the kinetic energy of particles [2]. Recently , ex- ploitation of Doppler effects was prop osed to amplify ul- trashort laser pulses [3, 4]. The Doppler effect also forms a practical constraint on the design of wireless comm uni- cations systems. With the gro wth of high-sp eed, highly- maneuv erable ob jects in the modern aerial environmen t, a thorough assessmen t of complex Doppler effects is crit- ical to the operation of electromagnetic infrastructure and the developmen t of further applications. Con tinu- ous c hanges in velocity create dynamic Doppler effects whic h app ear as chirps to observers. Nonlinear sensor motions induce new physical effects that cannot b e captured b y the lo cal constan t v elo cit y ap- pro ximation alone. T o provide a general corresp ondence b et w een the dynamics of mo ving ob jects and the chirps pro duced by them, we carefully derive the electromag- netic signals observed by dynamic non-inertial ob jects. W e utilize a geometric approac h based on the F renet- Serret frame of the observ er whic h can b e naturally de- scrib ed in 4D spacetime [5, 6]. The Doppler shift was also sho wn to b e fundamen tal to the Sagnac effect [7]. V ariable-frequency signals are im- p ortan t in man y areas of ph ysics suc h as pulsar detection and disp ersion measure searc hes [8]. As gravitational- w av e detectors b ecome more adv anced, metho ds are b e- ing developed for signals with fain t acceleration features [9]. In many applications, long-time coheren t integration can b e implemented to increase the signal-to-noise ratio ∗ Bryce.Barclay@asu.edu † Mahalov@asu.edu (SNR) [10], but this comes at the exp ense of introducing dynamic Doppler effects. Assessmen t of Doppler effects is increasingly critical to the operation of wireless com- m unications netw orks, such as LEO sate llite net works [11]. Although relativistic considerations can b e neglected in many engineering applications of the Doppler effect, relativit y has play ed an imp ortant role in the implemen- tation of many systems including the Global Positioning System [12, 13]. F urther, the ph ysical theory of electro- magnetic wa v es in non-inertial frames is well established in the literature [14–16], and it presents in teresting ph ys- ical phenomena which may b e exploited to dev elop new tec hnologies. In the developmen t of sophisticated space- time signal processing metho ds whic h utilize the Doppler effect, the 4D relativistic framework is adv antageous due to the Lorentz co v ariance of Maxwell’s equations. In previous work [17, 18], w e analyzed the sp ectra of signals receiv ed along nonlinear tra jectories in 3D space. Here, we presen t a detailed analysis of the dynamic Doppler effects of nonlinear relativistic motion in 4D spacetime, i.e., the effects of relativistic acceleration, jolt, and mov emen t in all 3 spatial dimensions on observed electromagnetic signals. Higher-order kinematic vectors suc h as jolt are relev ant for the study of mechanical sho c ks and for the control of acceleration (time-v arying acceleration) in engineering applications. T o this end, w e fix a stationary signal transmitter with the inertial co ordinate system ( x µ ) and describ e the kinematics of a non-inertial signal receiver in ( x µ ) co ordinates. In terms of the receiver’s w orld line z ( cτ ) = ( ct s ( cτ ) , x s ( ct ( cτ ))), where c is the sp eed of ligh t, x s ( ct ) is the receiver tra jec- tory in 3D space, and τ is the proper time of the receiv er, 2 the relativistic v elo cit y , acceleration, and jolt are [19, 20] u = ∂ z µ ∂ cτ ∂ ∂ x µ (1) a = ∂ u µ ∂ cτ ∂ ∂ x µ (2) σ = j − a µ a µ u, where j = ∂ a µ ∂ cτ ∂ ∂ x µ . (3) The electromagnetic tensor is the alternating cov ariant 2-tensor field: F µν =    0 − E x /c − E y /c − E z /c E x /c 0 B z − B y E y /c − B z 0 B x E z /c B y − B x 0    . (4) T o transform the electromagnetic field tensor to the p er- sp ectiv e of the receiver, construct the frame ( e µ ′ ) of the receiv er satisfying e µ ′ · e ν ′ = η µ ′ ν ′ , e 0 ′ = u, (5) d ( e λ ′ ) µ dcτ = ( u µ a ν − u ν a µ )( e λ ′ ) ν , (6) e µ ′ ( cτ ) = Λ ν µ ′ ( cτ ) e ν , (7) where ( e µ ) is the frame of the transmitter. The stationary transmitter with the co ordinates ( x µ ) that sends a contin uous-wa v e electromagnetic comm uni- cations signal given b y E y = E 0 exp( i 2 π f 0 ( ct − ˆ k ℓ x ℓ ) /c ) (8) B z = E 0 c exp( i 2 π f 0 ( ct − ˆ k ℓ x ℓ ) /c ) , (9) where the index ℓ indicates the spatial dimensions 1 , 2 , 3 only , i.e., ˆ k ℓ x ℓ = ˆ k 1 x 1 + ˆ k 2 x 2 + ˆ k 3 x 3 (otherwise, indices include the temporal dimension 0). The unit v ector ˆ k is the propagation direction and f 0 is the frequency of the wa v e. W e consider the microw a ve regime with f 0 = 1 GHz as a reference v alue. This w ork is organized as follo ws. In section I I, the sig- nal spectrum observ ed by a receiv er with constant prop er jolt and with constant prop er acceleration are analyzed. In section I I I, the sp ectrum is analyzed for non-inertial receiv ers using the F renet-Serret geometric framework. Finally , in section IV, we discuss our results and present concluding remarks. I I. SPECTRAL ANAL YSIS WITH RELA TIVISTIC KINEMA TIC P ARAMETERS In this section, we consider a transmitted signal prop- agating in the direction ˆ k = (1 , 0 , 0) and a receiver in the x 0 - x 1 plane with constant prop er jolt. The w orld line of the receiver is given by z 0 ( cτ ) = Z cτ 0 γ (cosh( ω ( cτ ′ )) + β sinh( ω ( cτ ′ ))) dcτ ′ (10) z 1 ( cτ ) = Z cτ 0 γ (sinh( ω ( cτ ′ )) + β cosh( ω ( cτ ′ ))) dcτ ′ + z 1 (0) . (11) Here, β = v 0 /c is the relative initial sp eed of the receiver in ( x µ ) co ordinates, γ = 1 / p 1 − β 2 , and ω ( cτ ) = a 0 cτ + j 0 ( cτ ) 2 / 2 where j 0 = | σ | and a 0 = | a (0) | . F or simplicit y , w e set z 1 (0) = 0. Along the tra jectory z ( cτ ), the receiver observes the signal given b y Eqs. (8)-(9). The receiv ed phase and w av en um b er functions are giv en by Φ s ( cτ ) = (12) 2 π f 0 c s 1 − β 1 + β Z cτ 0 exp( − ( a 0 cτ ′ + 1 2 j 0 ( cτ ′ ) 2 )) dcτ ′ ! K s ( cτ ) = d Φ s dcτ ( cτ ) = s 1 − β 1 + β k 0 exp( − ( a 0 cτ + 1 2 j 0 ( cτ ) 2 )) . (13) The received amplitude function is A s ( cτ ) = s 1 − β 1 + β exp( − ( a 0 cτ + 1 2 j 0 ( cτ ) 2 )) E 0 . (14) Setting a 0 = j 0 = 0, we reco ver the standard relativistic Doppler frequency and amplitude factor: D = s 1 − β 1 + β . (15) Because phase fluctuations occur on shorter timescales than amplitude fluctuations, w e use the stationary phase metho d (describ ed in the app endix) to appro ximate the sp ectrum of the signal: Z A ( x 0 ) exp( ik 0 h ( x 0 )) dx 0 ≈ (16) X m A ( x 0 m )  2 π − ik 0 h ′′ ( x 0 m )  1 / 2 exp( ik 0 h ( x 0 m )) , where x 0 m are the ro ots of h ′ ( x 0 ). T o obtain the spectrum of the receiv ed signal as a function of wa v enum ber k , w e use k 0 h ( x 0 ) = Φ s ( x 0 ) − k x 0 . The stationary phase 3 metho d gives Z exp( ik 0 h ( x 0 )) dx 0 ≈ (17) r − iπ k   2 j 0 log  Dk 0 exp( a 2 0 / 2 j 0 ) k    1 / 4 exp( ik 0 h ( x 0 + ))+ r iπ k   2 j 0 log  Dk 0 exp( a 2 0 / 2 j 0 ) k    1 / 4 exp( ik 0 h ( x 0 − )) , where the stationary p oints x 0 ± are x 0 ± = − a 0 j 0 ± v u u t 2 log  Dk 0 exp( a 2 0 / 2 j 0 ) k  j 0 . (18) In general, the contributions to the sp ectral approxima- tion (17) from the tw o stationary p oin ts create an inter- ference pattern in the sp ectrum. In the remainder of this section, we analyze the sp ec- trum of the receiv ed signal ov er time in terv als [ cτ i , cτ f ] where K s ( cτ ) is monotonic and only one stationary p oint con tribution is present, i.e., Z rect( x 0 ) exp( ik 0 h ( x 0 )) dx 0 ≈ (19) r − iπ k   2 j 0 log  Dk 0 exp( a 2 0 / 2 j 0 ) k    1 / 4 exp( ik 0 h ( x 0 + )) where rect( x 0 ) = 1 for cτ i ≤ x 0 ≤ cτ f and 0 otherwise. Using A = A s rect where A s is the received amplitude function giv en b y Eq. (14), the amplitude sp ectrum of the received signal is | S ( k ) | =      Z A ( x 0 ) exp( ik 0 h ( x 0 )) dx 0      ≈ (20) E 0   2 π 2 k 2 k 4 0 j 0 log  Dk 0 exp( a 2 0 / 2 j 0 ) k    1 / 4 . Substituting the wa v en umber function k = K s ( cτ ) giv es | S ( K s ( cτ )) | = (21) E 0  2 π 2 D 2 k 2 0 j 0  1 / 4 exp  − 1 2 ( a 0 cτ + 1 2 j 0 ( cτ ) 2 )   j 0 2 ( cτ ) 2 + a 0 cτ + a 2 0 2 j 0  1 / 4 . The ratio of the amplitude sp ectrum at cτ = cτ f to cτ = 0 is A j ( cτ f ) = | S ( K s ( cτ f )) | | S ( K s (0)) | = exp  − a 2 0 4 j 0  η 2 − 1   √ η , (22) 0.99995 0.99996 0.99997 0.99998 0.99999 1 1.00001 1.00002 10 0 0.99995 0.99996 0.99997 0.99998 0.99999 1 1.00001 1.00002 0.5 1 1.5 2 normalized amplitude 0.99995 0.99996 0.99997 0.99998 0.99999 1 1.00001 1.00002 normalized frequency 0.5 1 1.5 2 FIG. 1. Amplitude spectra of the signal received along the spacetime path z µ ( cτ ) giv en by Eqs (10)-(11) in blue. The frequency and amplitude are normalized by their resp ective v alues at cτ = 0. The sp ectrum appro ximation is obtained using the stationary phase metho d (20). In the top panel, prop er jolt j 0 and prop er acceleration a 0 are zero (classical Doppler shift). In the middle panel (chirp sp ectrum), prop er jolt j 0 is zero. In the b ottom panel (skew ed chirp sp ectrum), prop er jolt j 0 is nonzero. In the bottom tw o panels, the sta- tionary phase approximation is display ed in red. where η = j 0 a 0 cτ f + 1. The ratio of the signal w av en um b er at cτ = cτ f to cτ = 0 is D j = exp  − a 2 0 2 j 0  η 2 − 1   . (23) Then the amplitude factor is A j ( cτ f ) = s D j η . (24) When the proper jolt j 0 is zero, the w orld line of the receiv er is given b y Eqs. (10)-(11) where ω ( cτ ) = a 0 cτ and a 0 = | a (0) | . The received phase and w av en um b er functions are given b y Φ s ( cτ ) = − k 0 D a 0 exp( − a 0 cτ ) (25) K s ( cτ ) = d Φ s dcτ ( cτ ) = k 0 D exp( − a 0 cτ ) . (26) The ratio of the signal w av en um b er at cτ = cτ f to cτ = 0 is D a = exp( − a 0 cτ f ) . (27) The amplitude sp ectrum is given b y      Z exp( ik 0 h ( x 0 )) dx 0      ≈ r 2 π a 0 k = s 2 π exp( a 0 cτ ) a 0 k 0 D . (28) 4 0.99995 0.99996 0.99997 0.99998 0.99999 1 1.00001 1.00002 normalized frequency 10 0 normalized amplitude FIG. 2. Amplitude spectra of the signal received along the spacetime path z µ ( cτ ) given b y Eqs (10)-(11). The fre- quency and amplitude are normalized by their resp ective v al- ues at cτ = 0. The sp ectrum approximation is obtained us- ing the stationary phase metho d (20). The y ellow and blue curv es correspond to constan t proper acceleration and con- stan t proper jolt, resp ectiv ely . F or constant proper accelera- tion, the amplitude sp ectrum is nearly constant. The ampli- tude sp ectrum is strongly nonlinear for a receiver with con- stan t prop er jolt. The ratio of the amplitude at cτ = cτ f and cτ = 0 is A a ( cτ f ) = exp( − a 0 cτ f / 2) = p D a . (29) The amplitude sp ectra for constan t prop er jolt and con- stan t prop er acceleration are compared in Figs. 1 and 2. In Fig. 2, the green p oin t can b e mapp ed to the pur- ple p oint via the transformation ( k , S ) 7→ ( D a k , A a S ) or mapp ed to the red p oint via the transformation ( k , S ) 7→ ( D j k , A j S ). The jolt of a receiver accelerating aw a y from the signal emitter induces a nonlinear, sk ew ed chirp where the amplitude deca ys as the frequency shifts do wn. F or constan t prop er acceleration, the amplitude is nearly constan t as a function of frequency . I II. SPECTRAL ANAL YSIS WITH THE FRENET-SERRET FRAME IN 4D SP A CETIME The receiv er path can b e approximated by a series ex- pansion in terms of the geometric parameters of curv a- ture, torsion, and hyper-torsion of the 4D path using the F renet-Serret frame: z ( cτ ) = z (0) + z ′ (0) cτ + z ′′ (0) 2 ( cτ ) 2 + z ′′′ (0) 6 ( cτ ) 3 + z (4) (0) 24 ( cτ ) 4 + . . . = z (0) +  cτ + κ 2 1 6 ( cτ ) 3 + 3 κ 1 κ ′ 1 24 ( cτ ) 4  e 0 ′′ +  κ 1 2 ( cτ ) 2 + κ ′ 1 6 ( cτ ) 3 + κ 3 1 + κ ′′ 1 − κ 1 κ 2 2 24 ( cτ ) 4  e 1 ′′ +  κ 1 κ 2 6 ( cτ ) 3 + (2 κ ′ 1 κ 2 + κ 1 κ ′ 2 ) 24 ( cτ ) 4  e 2 ′′ +  κ 1 κ 2 κ 3 24 ( cτ ) 4  e 3 ′′ + . . . , (30) where deriv atives are tak en with resp ect to cτ and the parameters κ j and v ectors e k ′′ are ev aluated at cτ = 0. The F renet-Serret v ectors are given in 4D spacetime by de 0 ′′ dcτ = κ 1 e 1 ′′ (31) de 1 ′′ dcτ = κ 1 e 0 ′′ + κ 2 e 2 ′′ (32) de 2 ′′ dcτ = − κ 2 e 1 ′′ + κ 3 e 3 ′′ (33) de 3 ′′ dcτ = − κ 3 e 2 ′′ , (34) where e 0 ′′ corresp onds to the 4-v elo cit y and is tangent to the path, and the three spacelik e vectors are called the normal, binormal, and trinormal frame vectors. The geometric parameters κ 1 , κ 2 , and κ 3 are the curv ature, torsion, and hyper-torsion of the receiver tra jectory . T runcating to third order gives the approximate path z ( cτ ) ≈ z (0) +  cτ + κ 2 1 6 ( cτ ) 3  e 0 ′′ +  κ 1 2 ( cτ ) 2 + κ ′ 1 6 ( cτ ) 3  e 1 ′′ +  κ 1 κ 2 6 ( cτ ) 3  e 2 ′′ . (35) F or simplicit y , we set z (0) = 0. The 3-velocity of the observ er in the frame of the signal transmitter is given b y v ( cτ ) = cβ ( cτ ) = c dx s /dcτ dct/dcτ . (36) The received phase function for the emitted plane wa v e is Φ s ( cτ ) = k 0 k µ z µ ( cτ ) = k 0 k µ ζ ν ′′ ( cτ ) e ν ′′ µ ( cτ ) = k 0 ζ ν ′′ ( cτ ) α ν ′′ ( cτ ) (37) where k µ = (1 , − ˆ k 1 , − ˆ k 2 , − ˆ k 3 ) and ζ ν ′′ ( cτ ) are the co ef- ficien ts of z ( cτ ) with resp ect to the e ν ′′ frame. Here w e 5 define α ν ′′ ( cτ ) = k µ e ν ′′ µ ( cτ ) and w µ = ( κ 2 1 , κ ′ 1 , κ 1 κ 2 , 0). Using the series expansion (30), the received phase is Φ s ( cτ ) = k 0 α ν ′′ (0) ζ ν ′′ ( cτ ) (38) and the received w av en um b er function is K s ( cτ ) = k 0 α ν ′′ (0) dζ ν ′′ dcτ ( cτ ) = k 0  α µ w µ ( cτ ) 2 2 + α 1 κ 1 cτ + α 0  . (39) Because the w a v enum ber function is not necessar- ily monotonic, we consider the monotonic and non- monotonic cases separately , b eginning with the mono- tonic case. F or the third-order observer path expansion, the stationary phase metho d gives Z exp( ik 0 h ( x 0 )) dx 0 ≈ (40) √ − 2 π i exp( ik 0 h ( x 0 + )) ( k 0 ( α 2 1 κ 2 1 k 0 +2( α µ w µ )( k − α 0 k 0 ))) 1 / 4 + √ 2 π i exp( ik 0 h ( x 0 − )) ( k 0 ( α 2 1 κ 2 1 k 0 +2( α µ w µ )( k − α 0 k 0 ))) 1 / 4 . F urther, h 0 = h ( x 0 + ) + h ( x 0 − ) 2 (41) = α 1 κ 1 ( α 2 1 κ 2 1 k 0 + 3( α µ w µ )( k − α 0 k 0 )) 3 k 0 ( α µ w µ ) 2 h 1 = h ( x 0 + ) − h ( x 0 − ) 2 (42) = ( α 2 1 κ 2 1 k 0 + 2( α µ w µ )( k − α 0 k 0 )) 3 / 2 3 k 3 / 2 0 ( α µ w µ ) 2 . F or a monotonic wa v en umber function K s ( cτ ), the sp ec- trum approximation consists of one stationary p oint: Z rect( x 0 ) exp( ik 0 h ( x 0 )) dx 0 ≈ (43) √ − 2 π i exp( ik 0 h ( x 0 + )) ( k 0 ( α 2 1 κ 2 1 k 0 +2( α µ w µ )( k − α 0 k 0 ))) 1 / 4 . The approximation giv en b y Eq. (43) is display ed in Fig. 3. The result of acceleration along a nonlinear spatial tra jectory x s ( ct ) is a nonlinear, sk ewed chirp with an amplitude decay as the frequency shifts down similar to the chirp induced by jolt discussed in section I I. The ratio of the amplitude at cτ = cτ f and cτ = 0 is A F S ( cτ f ) = 1 √ η , (44) where η = α µ w µ α 1 κ 1 cτ f + 1. The ratio of the signal w av en um- b er at cτ = cτ f and cτ = 0 is D F S = ( α 1 κ 1 ) 2 2 α 0 ( α µ w µ )  η 2 − 1  + 1 . (45) 0.99995 0.99996 0.99997 0.99998 0.99999 1 1.00001 1.00002 1.00003 1.00004 normalized frequency 10 -2 10 -1 10 0 normalized amplitude FIG. 3. Comparison of the sp ectrum of the signal receiv ed along the spacetime path z µ ( cτ ) given by Eq. (30) (in blue) to the sp ectrum approximation (40) obtained from the sta- tionary phase metho d (in red). The receiver path z µ ( cτ ) is describ ed by the 4D F renet-Serret expansion. The wa v enum ber function is non-monotonic around the critical v alue k c : k c = k 0  α 0 − ( α 1 κ 1 ) 2 2 α µ w µ  . (46) T o provide a uniform approximation for the non- monotonic case, the Airy function appro ximation is used (see the app endix): 2 π exp( ik 0 h 0 ) ( α µ w µ k 0 / 2) 1 / 3 Ai −  3 2 k 0 h 1  2 / 3 ! . (47) The amplitude sp ectrum for this scenario is displa yed in Fig. 4. Because the frequency of the received sig- nal is a non-monotonic function of time, differen t times corresp onding to the same frequency interfere, creating amplitude fluctuations in the sp ectrum. IV. CONCLUSION In this work, we presented a thorough analysis of the sp ectral artifacts of non-inertial observ er motion. W e de- riv ed in terpretable expressions for the dynamic Doppler shift and sp ectral amplitude growth (or decay). As further Doppler-based tec hnologies are established, pre- cise quan titative descriptions of the nonlinear sp ectral features of emitted and observed signals will b e neces- sary . These sp ectral features can b e utilized to dev elop no vel physics-informed signal pro cessing, AI, and ma- c hine learning algorithms for Doppler mitigation and/or exploitation strategies in next-generation comm unica- tions, sensing, and radar applications. This w ork estab- lishes building blo c ks from which general motions in 4D 6 1 1.0000005 1.000001 1.0000015 1.000002 normalized frequency 0 0.2 0.4 0.6 0.8 1 normalized amplitude FIG. 4. The Airy approximation (47) of the sp ectrum of the signal received along the 4D F renet-Serret path z µ ( cτ ) giv en by Eq. (30). The received signal frequency is a non- monotonic function of time creating an interference pattern in the sp ectrum. spacetime can b e delineated. F urther developmen t in- cludes the analysis of Bez ´ ıer curves in 4D spacetime and the creation of algorithms to detect dynamic Doppler ef- fects induced by motion along 4D Bez ´ ıer curv es. A CKNOWLEDGMENTS This material is based upon work supp orted b y the Air F orce Office of Scientific Researc h under aw ard num ber F A9550-23-1-0177. [1] K. Abratkiewicz, M. Malanowski, and Z. Ga jo, T arget ac- celeration estimation in active and passive radars, IEEE J. Sel. T op. Appl. Earth Obs. Remote Sens. 16 , 9193 (2023). [2] J. Kw olek and A. Black, Contin uous sub-doppler-co oled atomic b eam interferometer for inertial sensing, Physical Review Applied 17 , 024061 (2022). [3] K. De Kinder, A. Bahrami, and C. Caloz, Doppler pulse amplification, Physical Review Applied 24 , 044029 (2025). [4] K. De Kinder, A. Bahrami, and C. Caloz, Scattering and c hirping at accelerated interfaces, Ph ysical Review Ap- plied 25 , 014077 (2026). [5] K. Hari and D. Kothaw ala, Rotating detectors in dS and AdS spacetimes, Physical Review D 109 , 104073 (2024). [6] B. M. Barclay and A. Mahalo v, Spectral distortions of signals induced b y nonlinear sensor motion: Acceler- ation, jolt, and relativistic effects, IEEE 59th Annual Asilomar Conference on Signals, Systems and Comput- ers: Signal Pro cessing and Learning for Communication Systems (2025. T o b e published in IEEE Xplore, Marc h 2026.). [7] J. C. How ell, M. Kahn, E. Grynszpan, Z. R. Cohen, S. Residori, and U. Bortolozzo, Doppler gyroscop es: fre- quency vs phase estimation, Physical Review Letters 129 , 113901 (2022). [8] F. Jenet and T. Prince, Detection of v ariable frequency signals using a fast chirp transform, Physical Review D 62 , 122001 (2000). [9] K. Chamberlain, C. J. Moore, D. Gerosa, and N. Y unes, F requency-domain w a v eform appro ximan ts capturing doppler shifts, Physical Review D 99 , 024025 (2019). [10] P . Huang, X.-G. Xia, G. Liao, Z. Y ang, and Y. Zhang, Long-time coherent integration algorithm for radar ma- neuv ering weak target with acceleration rate, IEEE T rans. Geosci. Remote Sens. 57 , 3528 (2019). [11] B. M. Barclay , E. J. Kostelic h, and A. Mahalov, Sensor placemen t sensitivity and robust reconstruction of wa v e dynamics from m ultiple sensors, SIAM Journal on Ap- plied Dynamical Systems 21 , 2297 (2022). [12] N. Ashb y , Relativity in the Global Positioning System, Living Rev. Relativ. 6 , 1 (2003). [13] G. Krieger and F. De Zan, Relativistic effects in bistatic syn thetic ap erture radar, IEEE T rans. Geosci. Remote Sens. 52 , 1480 (2013). [14] R. Neutze and G. Stedman, Detecting the effects of linear acceleration on the optical resp onse of matter, Ph ysical Review A 58 , 82 (1998). [15] J. C. Hauck and B. Mashhoon, Electromagnetic wa v es in a rotating frame of reference, Annalen der Physik 515 , 275 (2003). [16] J. V an Bladel, R elativity and engine ering , V ol. 15 (Springer-V erlag, 1984). [17] B. M. Barclay , E. J. Kostelich, and A. Mahalov, Ph ysics- informed signal pro cessing for time series data from ac- celerating sensors, IEEE T rans. An tennas Propag. 73 , 528 (2025). [18] B. M. Barclay , E. J. Kostelich, and A. Mahalov, Doppler effects of nonlinear sensor motion in 3D space: Curv a- ture, torsion, jolts, and directional wa v e propagation, IEEE T rans. Antennas Propag. 73 , 1840 (2025). [19] J. G. Russo and P . K. T ownsend, Relativistic kinematics and stationary motions, J. Ph ys. A: Math. Theor. 42 , 445402 (2009). [20] J. M. Pons and F. de P alol, Observ ers with constant prop er acceleration, constant prop er jerk, and b ey ond, Gen. Relat. Gravit. 51 , 80 (2019). [21] A. Erd´ elyi, Asymptotic representations of Fourier in te- grals and the metho d of stationary phase, Journal of the So ciet y for Industrial and Applied Mathematics 3 , 17 (1955). 7 [22] E. M. Stein, Harmonic analysis: r e al-variable methods, ortho gonality, and oscil latory inte gr als , V ol. 3 (Princeton Univ ersity Press, 1993). App endix A: The method of stationary phase and the Airy framework The stationary-phase appro ximation of an oscillatory in tegral is given by Z A ( t ) exp( iλh ( t )) dt ≈ (A1) X m A ( t m )  2 π − iλh ′′ ( t m )  1 / 2 exp( iλh ( t m )) , where the stationary times t m are the ro ots of h ′ ( t ). W e now provide the stationary-phase approximation of an oscillatory integral on a finite time interv al [ T 1 , T 2 ]: Z T 2 T 1 exp( iλh ( t )) dt, (A2) with a general, cubic phase h ( t ) = at 3 + bt 2 + ct , where a > 0, b , and c are any fixed constants. Integration o ver the finite in terv al [ T 1 , T 2 ] is represen ted by setting A ( t ) = 1 for t in [ T 1 , T 2 ] and setting A ( t ) = 0 otherwise in (A1). The approximation can be divided into three cases dep ending on the num ber and m ultiplicity of the stationary p oin ts contained in the in terv al of integration [ T 1 , T 2 ]. The stationary p oints of h ( t ) are the ro ots of h ′ ( t ), denoted by t ± . The first case is when T 1 < t − < t + < T 2 . The stationary-phase appro ximation is Z T 2 T 1 exp( iλh ( t )) dt ≈ π 1 / 2 exp( i ( λh ( t − ) − π / 4)) [ b 2 − 3 ac ] 1 / 4 λ − 1 / 2 + π 1 / 2 exp( i ( λh ( t + ) + π / 4)) [ b 2 − 3 ac ] 1 / 4 λ − 1 / 2 , where the b oundary terms of order λ − 1 : exp( i ( λh ( T 1 ) + π / 2)) 3 a ( t − − T 1 )( t + − T 1 ) λ − 1 + exp( i ( λh ( T 2 ) − π / 2)) 3 a ( T 2 − t − )( T 2 − t + ) λ − 1 are omitted. F or the cubic phase h ( t ), the lo cal extrema h ( t ± ) can b e written explicitly: h ( t ± ) = 2 b 3 − 9 abc ∓ 2[ b 2 − 3 ac ] 3 / 2 27 a 2 = h 0 ∓ h 1 . (A3) Then the stationary phase approximation is Z T 2 T 1 exp( iλh ( t )) dt ≈ 2 π 1 / 2 (3 a ) 1 / 3 cos( λh 1 − π / 4) exp( iλh 0 )  3 2 h 1  1 / 6 λ − 1 / 2 . See [21] and [22], c h. 8 for more information on the stationary-phase metho d. The second case is when T 1 < t − = t + < T 2 . It is useful in this case to provide an appro ximation which is uniformly v alid when t + − t − is near zero. T o accomplish this, the integral (A2) is ap- pro ximated b y Airy functions using the transformation: h ( t ) = h 0 −  3 2 h 1  2 / 3 s + 1 3 s 3 , (A4) where dt ds = s 2 −  3 2 h 1  2 / 3 3 at 2 + 2 bt + c (A5) = ∞ X m =0 p m s 2 −  3 2 h 1  2 / 3 ! m + q m s s 2 −  3 2 h 1  2 / 3 ! m , giving Z T 2 T 1 exp( iλh ( t )) dt ≈ 2 π exp( iλh 0 ) (3 aλ ) 1 / 3 Ai −  3 2 λh 1  2 / 3 ! . (A6) The third case is when only one ro ot is in the interv al, whic h for simplicity we will assume is t − . The stationary- phase approximation in this case is Z T 2 T 1 exp( iλh ( t )) dt ≈ π 1 / 2 (3 a ) 1 / 3 exp( i ( λ ( h 0 + h 1 ) − π / 4))  3 2 h 1  1 / 6 λ − 1 / 2 .