Underwater Backscatter Localization: Toward A Battery-Free . - MIT
Underwater Backscatter Localization:Toward a Battery-Free Underwater GPSReza Ghaffarivardavagh, Sayed Saad Afzal, Osvy Rodriguez, and Fadel AdibMassachusetts Institute of RACTCan we build a battery-free underwater GPS? While underwater localization is a long-studied problem, in this paper, we seek to bringit to battery-free underwater networks [13, 20]. These recentlyintroduced networks communicate by simply backscattering (i.e.,reflecting) acoustic signals. While such backscatter-based communication enables them to operate at net-zero power, it also introducesnew and unique challenges for underwater localization.We present the design and demonstration of the first underwaterbackscatter localization (UBL) system. Our design explores various challenges for bringing localization to underwater backscatter,including extreme multipath, acoustic delay spread, and mobility.We describe how an adaptive and context-aware algorithm mayaddress some of these challenges and adapt to diverse underwaterenvironments (such as deep vs shallow water, and high vs low mobility). We also present a prototype implementation and evaluationof UBL in the Charles River in Boston, and highlight open problemsand opportunities for underwater backscatter localization in oceanexploration, marine-life sensing, and robotics.CCS CONCEPTS Networks Network architectures; Hardware Wireless integrated network sensors; Applied computing Environmental sciences.KEYWORDSSubsea IoT, GPS, Localization, Backscatter Communication, Piezoelectricity, Wireless, Energy Harvesting, Battery-free, AcousticsACM Reference Format:Reza Ghaffarivardavagh, Sayed Saad Afzal, Osvy Rodriguez, and Fadel Adib.2020. Underwater Backscatter Localization: Toward a Battery-Free Underwater GPS. In Proceedings of the 19th ACM Workshop on Hot Topics in Networks(HotNets ’20), November 4–6, 2020, Virtual Event, USA. ACM, New York, NY,USA, 7 pages. ONThere is significant interest in low-power and distributed underwater localization systems for environmental, industrial, and defenseapplications [9, 22, 28, 38]. Climatologists and oceanographers areinterested in deploying such systems to obtain location-taggedPermission to make digital or hard copies of part or all of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full citationon the first page. Copyrights for third-party components of this work must be honored.For all other uses, contact the owner/author(s).HotNets ’20, November 4–6, 2020, Virtual Event, USA 2020 Copyright held by the owner/author(s).ACM ISBN 22604.3425950ocean measurements for constructing subsea heatmaps [28], understanding ocean processes [33], and developing accurate weatherand climate prediction models [32]. Marine biologists are interestedin such systems for tracking schools of fish and studying theirbehavior and migration patterns [7, 22]. Accurate and low-powerlocalization is also a key enabler for various underwater robotictasks including navigation, tagging, and object manipulation [9, 27].Unfortunately, prior designs for underwater localization remainfar from the vision of a low-power, low-cost, and scalable architecture. Since standard GPS signals do not work in water,1most existing underwater positioning systems rely on acousticsignals [5, 6, 24]. These systems typically require their nodes torepeatedly transmit acoustic beacons (which are used by a remote receiver for triangulation). Such repetitive transmissions canquickly drain a sensor’s battery, thus requiring a frequent andexpensive process of battery replacement [10, 11]. To avoid thisproblem, existing localization systems either heavily duty-cycletheir transmissions [12, 42] or tether the localization beacons to alarge power source on a ship or submarine [19, 23]. Unfortunately,such workarounds prevent these systems from accurately trackingfast-moving objects (like fish or drones) and/or scaling to largeareas of the ocean.We introduce Underwater Backscatter Localization (UBL), anultra-low power and scalable system for underwater positioning.UBL builds on our recent work in underwater networking, whichhas demonstrated the potential to communicate at near-zero powervia acoustic backscatter [13, 20]. In contrast to traditional underwater acoustic communication systems, which require each sensor togenerate its own signals, backscatter nodes communicate by simplyreflecting acoustic signals in the environment. These nodes canalso power up by harvesting energy from acoustic signals. Thus, bybringing localization to underwater backscatter, UBL would enableus to build a long-lasting, scalable, battery-free underwater GPS.Before explaining how UBL works, let us understand why itcannot easily adopt traditional underwater localization techniques.State-of-the-art underwater localization systems rely on computing the time-of-arrival (ToA) between two nodes [4, 43].2 In thesesystems, a transceiver sends out an acoustic pulse, and waits for aresponse from the transponder beacon. The time difference betweenthe initial pulse and the reply is used to determine the separationbetween the two nodes (by multiplying it with the sound speed inwater). Unfortunately, this ToA estimation technique does not workfor battery-free nodes. These nodes require an additional wake-uptime to harvest energy from acoustic signals before they can startbackscattering. This wake-up time cannot be determined a priori1 GPSrelies on RF signals which decay exponentially underwater [31, 36].contrast, using angle-of-arrival typically requires expensive and bulky antennaarrays and results in poorer accuracy than ToA [16].2 In
and varies with location and environment. As a result, it adds anunknown offset to the time difference between the transmitted andreceived pulse, preventing us from accurately estimating the ToAand using it for localization.3To overcome this challenge, UBL adopts a time-frequency approach to estimate the ToA. Specifically, instead of estimatingthe ToA entirely in the time domain, it also collects frequencydomain features by performing frequency hopping. Since timeand frequency are inversely proportional, hopping over a widebandwidth would enable UBL to estimate the ToA with highresolution [1, 26, 40]. Transforming this idea into a practical underwater localization system still requires dealing with multipleconfounding factors: Multi-path: When acoustic signals are transmitted underwater,they repeatedly bounce back and forth between the seabed andthe water surface before arriving at a receiver. Such dense multipath reflections make it difficult to isolate the direct path to abackscatter node for ToA estimation. Delay Spread: The slow speed of sound propagation spreadsout the above multipath reflections over time, resulting in a largedelay spread. This delay spread causes different backscatter bits– even from the same node – to interfere with each other. As weshow in §2, such inter-backscatter interference is unique to acoustic backscatter and exacerbates the ToA estimation problem.4 Mobility: Performing accurate localization becomes more challenging for mobile nodes (fish, drones). This is because mobility distorts the estimated frequency features (due to Dopplershift [39]) and because frequency hopping increases the latencyof localization, during which a mobile node may have moved toa new location.Addressing the above challenges simultaneously requires satisfying competing design requirements. For example, reducing thebackscatter bitrate would increase the separation between symbolsin a packet (thus mitigating inter-symbol interference), but it alsoslows down the channel estimation process, making it difficult tolocalize fast-moving objects. In a similar vein, dealing with multipath and mobility results in conflicting design constraints (for thebitrate and hopping sequence). We argue that designing a robustunderwater backscatter localization system requires context-awarealgorithms that can adapt their bitrates and hopping sequence totheir operating domains. In §3.2, we describe the fundamental constraints arising from these different challenges and how our designof UBL aims to strategically adapt to its surrounding environment.We implemented a proof-of-concept prototype for UBL andtested in a river. Our prototype consists of a mechanically fabricated backscatter node and a custom-made PCB with a microcontroller and backscatter logic. Our experimental evaluation acrossthree different locations demonstrates the feasibility of achievingcentimeter-level accuracy using UBL. Our empirical evaluation iscomplemented with simulations that demonstrate how the systemcan adapt to different speeds and multipath environments.Contributions: This paper presents the first design and demonstration of underwater backscatter localization. Our design can deal3 Note that an approach that introduces pauses between a reader’s transmissions is alsoundesirable since the backscatter node requires continuous signals to stay awake [20].contrast, in RF backscatter localization, due to the high propagation speed, allmultipath reflections arrive in the same backscatter state [25, 26].4 In(a) Multipath in underwater channel(b) Backscatter signal in deep and shallow waterFigure 1—Multipath and Underwater Backscatter. (a) shows how sound propagates underwater, repeatedly reflecting off the surface and seabed. (b) shows a receivedbackscatter packet in deep (low multipath) and shallow (dense multipath) water.with unique challenges that arise from the interaction between underwater multipath and acoustic backscatter, and it can can adaptto various underwater conditions (depth, mobility). The paper alsocontributes a proof-of-concept implementation and evaluation, andit highlights open problems and future opportunities in underwaterbackscatter localization.2THE (NEW) PROBLEMBefore describing UBL’s design, it is helpful to understand whyunderwater backscatter localization poses new challenges that aredifferent from prior work in RF backscatter localization (e.g., RFIDlocalization [14, 25, 26, 41]). To answer this question, in this section,we provide background on underwater acoustic channels, thenexplain how these channels pose interesting new challenges forbackscatter localization.Underwater Acoustic Channel. The underwater channel is aconfined environment bounded with air on one side and sedimenton the other side as shown in Fig. 1(a). When acoustic signals aretransmitted underwater, they can travel over very long distances(tens to hundreds of kilometers [34]) due to two factors: (1) thesmall attenuation of sound in water; and (2) the fact that soundentirely reflects off the air/water and water/sediment boundariesbecause of the large impedance mismatch between these media.Thus, an acoustic signal travels on various paths from a transmitterto a receiver, most of which involve multiple reflections off the airand water boundaries. As a result, the receiver obtains multiplecopies of the signal, which we refer to as underwater multi-path.Impact of Multipath on Underwater Backscatter. To understand the impact of the underwater channel on acoustic backscatter,we simulated backscatter communication in two different environments corresponding to deep water (depth 200m) and shallow
water (depth 10m). In both of these environments, the backscatternode and the receiver are separated by the same distance (4 m).Fig. 1(b) shows the received backscatter signal in each of thesetwo scenarios. In deep water (black plot of Fig. 1(b)), the receivedsignal shows clear transitions between reflective and non-reflectivestates. Recall that these states encode bits of 0’s and 1’s that areused to communicate data. In contrast, in shallow water (orangeplot of Fig. 1(b)), the backscatter response is highly distorted andthe transitions are significantly obscured.5 It is worth noting thatthe difference between these two scenarios is not due to differencein the signal-to-noise ratio, since the distance separation betweenthe backscatter node and the receiver is the same in both cases.Instead, the difference between the two different scenarios arisesfrom the multipath reflections mentioned earlier. Specifically, indeep water, the direct path is much stronger than the reflectedpaths because it travels a smaller distance and experiences lessattenuation (4m vs 200m). In contrast, in shallow water, the directpath and reflected paths have similar lengths and thus have similaramplitudes; this leads to interference between subsequent symbols(i.e., between different backscatter states). Unless this distortion isaccounted for, it will be difficult to estimate the wireless channel inthe frequency domain (which UBL needs for localization).This inter-symbol interference (ISI) is unique to underwaterbackscatter and does not exhibit in RF backscatter.6 The differencebetween RF and acoustic backscatter arises from significant disparity between the speed of RF in air (3 108 m/s) and that of sound inwater (1, 500m/s). In RF backscatter, the nearest reflector that maycause ISI is more than 3 km away (i.e., significantly attenuated),while in acoustic backscatter even a reflector that is 1.5 m away cancause ISI. This difference motivates a new principled approach forunderwater localization that differs from standard RF backscatterlocalization techniques.3UBLUBL is an accurate underwater localization system for ultra-lowpower and battery-free nodes. The system can achieve centimeterscale positioning even in multipath-rich underwater environments.To locate a backscatter sensor, UBL performs the following steps: A UBL reader sends a query searching for backscatter nodes inthe environment. When a node replies, UBL sends a downlink commands specifying the backscatter bitrate. As the node replies, the reader performs frequency hopping toestimate the node’s channel over a wide frequency. Finally, UBL uses the acquired bandwidth to estimate the timeof-arrival (ToA) to the backscatter node and uses the ToA forlocalization.Since prior work has demonstrated the ability to query andcommand an underwater backscatter node [20], in this section, wefocus on how UBL uses frequency hopping to estimate the ToA(§3.1) and how it selects the bitrate and hopping sequence (§3.2).5 Weobserved similar behavior when empirically testing our system in a real river.3.1Backscatter ToA EstimationUBL performs localization by estimating the time-of-arrival (ToA)of a backscatter node’s signal. ToA estimation is particularly usefulin multipath-rich environments. Specifically, in the presence ofmultiple reflections, a receiver can determine the direct path as theone having smallest ToA (since it travels along the shortest path).The main challenge in backscatter ToA estimation arises fromthe random wake-up time of battery-free nodes. Specifically, recallfrom §1 that battery-free nodes need to harvest energy in order topower up before they can start backscattering. Moreover, this wakeup delay varies with distance and environment; thus, it cannot bedetermined a priori.To overcome this challenge, instead of estimating ToA directlyin the time domain, UBL does so in the frequency domain. Sincetime and frequency are inversely related, a wide bandwidth can beused to separate different paths and identify the direct path as theone that arrives earliest. Specifically, the resolution to determinethe direct path is given by the following ���𝑑𝑡ℎThus, for a bandwidth of 10kHz, UBL can localize the node to within10 cm. In the rest of this section, we describe the different steps ofUBL’s ToA estimation approach.7resolution Stage 1: Wideband Channel Estimation. UBL estimates thebackscatter channel over a wide bandwidth by performing frequency hopping. Specifically, it transmits a downlink signal at afrequency 𝑘 and obtains the backscatter response. Once the receiverobtains the response 𝑦𝑡 , it performs the following two steps:(1) First, it cross-correlates the received signal with the knownbackscatter packet preamble to determine the beginning ofa packet, denoted 𝜏 , using theÕ following equation: 𝜏 arg max𝑦𝑡 𝜏 𝑝𝑡(1)𝜏𝑡 𝑇where 𝑝𝑡 is the known preamble and 𝑇 is the length of thepreamble. By identifying the beginning of the packet, thiscorrelation can be used to eliminate the wake-up lag.(2) Subsequently, UBL estimates the backscatter channel 𝐻𝑘using the packet’s preamble. This can be done using standardchannel estimation as per the following equation:1Õ𝑦𝑡 𝜏 𝑝𝑡(2)𝐻𝑘 𝑇𝑡 𝑇where 𝑦𝑡 𝜏 corresponds to the received signal shifted to thebeginning of a packet.UBL repeats the above procedure for different frequencies (eachtime hopping to a different frequency and computing the corresponding channel) until is has obtained the channels for across awide bandwidth [𝐻 1, 𝐻 2, . . . 𝐻 𝑁 ].Stage 2: Obtaining the Time-Domain Channel. After concatenating the different frequencies, UBL performs an inverse Fouriertransform (IFFT) on the channels. This allows it to obtain an expression of the channel in the time domain. Importantly, this timedomain representation is independent of the random wake-up timesince it is obtained entirely from the channel estimates.6 Note that ISI is known in wireless communication, and standard protocols like OFDMcan be used to address it [39]. However, OFDM is too complex for battery-free underwater nodes.7 Wenote that this technique is similar to that employed in [26] and will be adaptedin §3.2 to underwater backscatter.
Signal AmplitudeSignal AmplitudePeak(direct path)2Estimate Error (%)Conventional ToA EstimationFrequency-Hopping based ToA e (m)Distance (m)(a) Shallow water, Bit-rate: 2 kbits/s(b) Shallow water, Bit-rate: 100 bits/sFigure 3—ToA Estimation in shallow water. This figure shows how multipathaffects the localization ability for UBL. (a) shows that at a high bit-rate of 2 kbits/s,UBL fails to localize the object. (b) shows that operating at a lower bit-rate of 100 bits/sin multipath rich environments yields better performance.(a) Time-domain channel from frequency hopping6002.5400is robust to the random wake-up lags of battery-free backscattersensors. In §4, we empirically verify this result as well.3002003.2100000.0050.010.0150.02Wake-up lag (s)0.0250.03(b) Wake-up lag effect of localizationFigure 2—Range estimation via frequency hopping. (a) shows how UBL can isolate the direct path in the time domain using the frequency-hopping localizationmethod and (b) shows the effect of the wake-up lag on conventional ToA based localization schemes.One might wonder whether eliminating the wake-up lag wouldalso eliminate the impact of the round-trip delay on the channel estimates. In practice, this does not happen because UBL estimates thechannel in the frequency domain. To see why this is true, considera simple setup with a single line-of-sight path from the backscatternode to the receiver. Here, the baseband received signal 𝑦𝑡 can beexpressed as:𝑦𝑡 𝑒 𝑗2𝜋 𝑓𝑐 𝜏𝑟 𝑏 (𝑡 𝜏𝑟 𝜏𝑤 )(3)where 𝜏𝑟 and 𝜏𝑤 correspond to the round-trip delay and wakeup lagrespectively. By shifting the received signal in the time domain (by𝜏𝑟 𝜏𝑤 ), UBL eliminates the delays in the time domain but not theimpact of the round-trip delay on the frequency-domain channel.Hence, it is able to recover this delay upon performing an IFFT.To demonstrate this idea in practice, we simulated the localization problem where a UBL reader and a backscatter node wereseparated by 4 m in a deep underwater environment. Fig.2(a) plotsthe channel amplitude as a function of distance after performing theabove procedures. The plot demonstrates a clear peak amplitude inthe channel around 4 m, which is aligned with the actual distanceof backscatter node. Note that because the simulated environmentcorresponds to a deep sea where multipath is distance with respectto the line of sight, the plot does not show other peaks from nearbyreflections in the environment.Next, to investigate the effect of wake-up lag on UBL’s ToAestimation approach, we simulated localization after introducingdifferent time delays (betweem 0-30 ms), and compared the outcome of UBL with that of conventional time-domain methods forlocalization. Fig. 2(b) plots the percentage error in distance estimation as a function of the wake-up lag for both schemes. The figureshows that while UBL’s error remains small irrespective of thewake-up lag, conventional (time-based) ToA estimation systems aresignificantly affected by this delay and suffer from a large marginof error. This demonstrates that UBL’s ToA estimation approachAdaptive Backscatter LocalizationSo far, we have described how UBL can estimate the ToA robustlydespite a random wake-up lag. However, the above descriptionfocused on deep sea environments with little multipath. In thissection, we describe how UBL’s design can be extended to dealwith extreme multipath and mobility in underwater environments.3.2.1 Dealing with Extreme Multipath. To understand the impactof extreme multipath, we repeated the same simulation of as ourearlier experiment but this time in shallow water (depth of 4 m)rather than in deep water. Fig. 3(a) plots the signal amplitude as afunction distance. Unlike the previous experiment, we are unableto see a sharp peak around 4 m, making it difficult to robustlyestimate the time-of-arrival in extreme multipath environments.This is because inter-symbol interference (ISI) makes it difficultto obtain accurate channel estimates. This challenge can be seenvisually in the orange plot of Fig. 1(b).To mitigate the impact ISI on ToA estimation, UBL can commandthe backscatter node to lower its bitrate. Intuitively, doing so increases the separation between any two backscatter symbols, thusreducing the interference between the reflection of the former withthe direct path of the latter. From a communication perspective,reducing the bitrate results in a more narrowband channel, whichincreases robustness to frequency selectivity [39].To test this idea, we repeated the same simulation, but this timeat a bitrate of 100 bps instead of 2 kbps. Fig. 3(b) plots the resultingoutput for this experiment. The figure shows a much sharper peakaround 4 m than that obtained when the same experiment wasperformed at a higher bitrate. The figure also shows a second peakaround 4.5 m, which corresponds to the first (primary) multipathreflection off the surface and sediment. Note that both experimentsused the same bandwidth and are simulated at the same distance(i.e., the latter did not benefit from more resolution or higher SNR).Rather the difference is localization robustness arises from the lowerbitrate. Formally, we can prove the following lemma.Lemma 3.1. To ensure the inter-symbol interference from any singlepath is no larger than 𝑘 dB, the backscatter bitrate must be less than𝑐(100.05𝑘 1) 4𝑟To prove this lemma, let us denote the largest delay caused by thereflected path as 𝑇𝑟 and the delay caused by the direct path as 𝑇𝑙 . Wefurther assume that the power of the reflected path is attenuated
3.2.2 Dealing with Mobility. Next, we are interested in extending UBL to deal with mobility of underwater backscatter (e.g., intracking fish, AUVs). To understand the impact of mobility on localization, we simulated the localization process in deep water for anode moving at a speed of 0.3m/s. Fig. 4 plots the signal amplitudeas a function of distance. Unlike the earlier experiment in deepwater (i.e., Fig. 2(b)), we are unable to see a sharp peak around 4 m,making it difficult to robustly estimate the time-of-arrival in thepresence of mobility. This is because mobility causes a change in thechannel estimates over time. As a result, the resulting channel estimates [𝐻 1 (𝑡 1 ), 𝐻 2 (𝑡 2 ), . . . 𝐻 𝑁 (𝑡 𝑁 )] cannot be coherently combinedto obtain an accurate location estimate.To mitigate the impact of mobility on ToA estimation, UBL needsto reduce the overall time required for localization. This can be doneby commanding the backscatter node to increase its bitrate andthe reducing the number of frequencies in the frequency hoppingsequence. We can formalize the mobility constraint through thefollowing lemma.Lemma 3.2. To localize a mobile node moving with the speed of𝑣, backscatter and frequency hopping properties should satisfy thecondition of : 𝑁 𝑓𝑅𝐿𝑝 2𝑣𝐵𝑐 where 𝑅 is the backscatter bitrate, 𝑁 𝑓 isthe number of frequency in frequency hopping, 𝐿𝑝 is the bit lengthof the preamble, 𝑣 is the relative speed of the mobile node , 𝐵 is thebandwidth and 𝑐 denotes the speed of sound.Lemma 3.2 is derived considering the fact that to localize a mobilenode with the resolution of 𝑥, frequency hopping process must beaccomplished before the node get displaced more than 𝑥. Assumingthe backscatter bitrate of 𝑅 and preamble bit length of 𝐿𝑝 , the minimum required time to estimate the channel for each frequency is𝐿𝑝𝑅 and the minimum required time for frequency hopping duration𝑁 𝐿is 𝑓𝑅 𝑝 . To localize the node, The duration of frequency hoppingshould be less than the time it takes for the node move more than𝑥. This gives us the following relation:𝑁 𝑓 𝐿𝑝𝑥 (4)𝑅𝑣Additionally, the resolution 𝑥 is function of bandwidth and may𝑐 , completing the lemma.be written as 𝑥 2𝐵 To test the relationship, we repeat the same experiment as above,but this time with a backscatter bitrate of 10kbps. (Here, 𝐵 10𝐾𝐻𝑧,𝐿𝑝 20, 𝑁 𝑓 100, this requiring a minimum bitrate of 8kbpsaccording to the lemma). Fig. 3(b) plots the resulting output for thisexperiment. The figure shows a much sharper peak around 4 mthan that obtained when the same experiment was performed at8 Thiscomes from standard spherical loss 𝑃 20 log10 (1/𝑇 ) .Signal AmplitudeSignal Amplitudeby 𝑘 dB compared to the power of the direct path. This gives us thefollowing relation:8𝑇𝑟 100.05𝑘 𝑇𝑙Since 𝑇𝑙 corresponds to the round trip distance, it can be written asa function of the separation 𝑟 and the speed of sound 𝑐 as 𝑇𝑙 2𝑟 /𝑐.To ensure that any strong symbol reflection arrives before the nextsymbol is received, the symbol period 𝑇𝑠 (or bit period) should begreater than twice the largest delay 𝑇𝑑 , which gives an upper boundfor bitrate 𝑅:𝑐𝑅 (100.05𝑘 1) 4𝑟22.533.544.555.5622.533.544.555.5Distance (m)Distance (m)(a) Deep water, Bit-rate: 1 kbits/s(b) Deep water, Bit-rate: 10 kbits/s6Figure 4—ToA Estimation in deep water with mobility. This figure shows howUBL can adapt to mobility in deep water environments . (a) shows that at a bit-rate of1 kbits, UBL is unable to localize the object while the object is moving with a speed of0.3 m/s, so for better accuracy, it is desirable to operate at a higher bit-rates to dealwith mobility as shown in (b).a lower bitrate, demonstrating that UBL’s adaptation enables it toaccurately localize despite mobility.We make few additional remarks about how UBL chooses itsbackscatter bitrate and hopping sequence: Lowering the bandwidth (𝐵), decreases the resolution of localization. Therefore, UBL always tries to exploit the full bandwidthallowed by the backscatter node’s mechanical characteristics. Decreasing the bit length of the preamble (𝐿𝑝 ), leads to the lowerSNR. UBL utilize the preamble length of at least 20 bit to ensurethe channel is estimated reliably. The longest distance that can be localized is determined by thelength of the IFFT. Therefore, decreasing the 𝑁 𝑓 , limits the rangeof the localization4FEASIBILITY STUDYIn this section, we explain how we implemented and validatedthe feasibility of UBL for underwater localization. Similar to ourprior work on underwater backscatter [2, 13, 20], UBL’s implementation leverages a projector to transmit an acoustic signal on thedownlink, a backscatter node that decodes the downlink signal andtransmits a backscatter packet on the uplink, and a hydrophone(Omnidirectional Reson TC 4014 hydrophone [37]) that receivesand decodes the backscatter packets. The projector and backscatter node were fabricated in house from piezoceramic cylinders,following the procedure elaborated in our prior work [20].In our experiment, the projector was programmed to hop itscarrier frequency from 7.5 kHz to 15 kHz (at 75 Hz intervals, eachfor 6 seconds). This range of frequency is selected based on bandwidth of the backscatter node [20] and, subsequently, the expectedresolution is 10cm. To account for the effect of multipath in suchshallow environment, the backscatter bitrate of 100 bit/s is adopted.Notably, since the node was relatively stationary in the water, thebitrate of 100 bit/s was sufficient to estimate the channel. The received signal recorded by the hydrophone was then processed byfirst estimating the channel at each of the frequencies, and subsequently, the time-domain channel is computed to estimate ToA perour discussion in §3.1.The outputs of UBL for three different node locations are shownin Fig.5. The x-axis corresponds to distance and the y-axis represents the normalized time-domain channel amplitude. In this result,the ground truth is marked using red vertical lines and the peakamplitude for each distance is within 10 cm from the ground truth.
6Figure 5—Preliminary Results for UBL. The system was tested for three differentranges i.e. 24 cm, 34 cm and 44 cm respectively.Note that, due to the limited bandwidth , our resolution was 10 cmand to achieve finer precision, UBL can emulate a wider bandwidth.5RELATED WORKUnderwater localization dates back to the early 20𝑡ℎ century. Thefirst underwater positioning system was developed to search for amissing American nuclear submarine, the USS Thresher [3]. Sincethen, the vast majority of the underwater positioning systems haverelied on acoustic beaconing (since GPS and radio signals do notwor
underwater backscatter localization poses new challenges that are different from prior work in RF backscatter localization (e.g., RFID localization [14, 25, 26, 41]). To answer this question, in this section, we provide background on underwater acoustic channels, then explain how these channels pose interesting new challenges for
networking [1, 43, 48, 56, 87]. Backscatter sensors can wirelessly communicate at near-zero power by simply reflecting radio signals in the environment. In this paper, we investigate the ability to take backscatter networking to underwater environments. In particular, since wireless communication is the largest source of energy con-
es the major management issues that are key to localization success and serves as a useful reference as you evolve in your role as Localization Manager. We hope that it makes your job easier and furthers your ability to manage complex localization projects. If the Guide to Localization Management enables you to manage localiza-
Localization processes and best practices will be examined from the perspective of Web developers and translators, and with these considerations in mind, an online localization management tool called Localize1will be evaluated. The process of localization According to Miguel Jiménez-Crespo (2013, 29-31) in his study of Web localization, the
devices. The Intel WISP [5] and UMass Moo [27] are examples of backscatter-enabled sensor platforms. Despite the energy benefits of backscatter radios, ex-isting network stacks achieve only short communication range and low throughput. We make an empiric argu-ment these limitations are, in part, due to the design of the network stack.
the limited depth of underwater welding. Welding equipment transformed from manual welding to underwater automatic welding. The efficient and low-cost underwater welding was achieved[7]. In order to study the automatic welding technology under larger deep-water environment, the underwater automatic welding system was designed in this paper. The
cast caused by the effects of underwater imaging conditions deteriorate the capability to fully extract valuable information from underwater images for further processing such as marine, mine detection and aquatic robot inspection. Hence, it is of great interest to restore degraded underwater images for high-quality underwater imaging [3].
In the localization of any software including websites and web apps, mobile apps, games, IoT and standalone software, there is no continuous, logical document similar . Localization workflow best practices 04 Localization workflow. Lokalise is a multiplatform system — that means you can store iOS, Android, Web or
CALCULUS fall 2009 Typeset:June 8, 2010 1. MATH 221 { 1st SEMESTER CALCULUS LECTURE NOTES VERSION 2.0 (fall 2009) This is a self contained set of lecture notes for Math 221. The notes were written by Sigurd Angenent, starting from an extensive collection of notes and problems compiled by Joel Robbin. The LATEX and Python les which were used to produce these notes are available at the following .
