Abstract
Distance measuring equipment (DME) has been a cornerstone of aviation navigation for the last 70 years. While GNSS is taking an increasingly important role in civil aviation, DME can still play an important role in a robust aviation infrastructure in the foreseeable future. Advanced concepts have been developed to improve DME performance and capabilities. One concept is a DME-based pseudolite that position modulates existing DME pulse pairs. It is interoperable with DME operations and can be generated using the currently fielded DME transponders with an appliqué. This concept is suitable as a robust alternative to GNSS for aviation navigation or timing. The paper examines the performance of DME pseudolite implemented via an appliqué on a DME transponder. The paper examines the synchronization and data performance of the DME pseudolite signal in the air and on the ground. The paper compares the actual performance to theoretical results.
1 INTRODUCTION
Distance measuring equipment (DME) and its military counterpart tactical air navigation (TACAN) have been a cornerstone of aviation navigation since their development and introduction in the 1940s and 1950s. It is still one of the most commonly used aviation navigation aids. However, future airspace will need to accommodate increased traffic levels, more precise operations, and unmanned aerial vehicles (UAVs), which will increase demands on aviation navigation systems. While global navigation satellite systems (GNSS) will provide many of these improvements, terrestrial navigation systems will continue to play an important role in a robust aviation navigation infrastructure.
The Federal Aviation Administration (FAA) and other organizations have examined advanced concepts to improve on DME performance and capabilities while maintaining compatibility with existing equipment. One concept being developed is a DME-based pseudolite that position modulates existing DME pulse pairs. Such a pseudolite can operate alongside current DME operations and can be generated using the currently fielded DME transponders with an external appliqué. This concept could provide ranging and timing service thus offering an attractive source of alternative position, navigation, and timing (APNT) to GNSS (Eldredge et al., 2010).
This paper demonstrates the implementation and on-air performance of the DME pseudolite appliqué on a DME transponder. Specifically, it details the design and implementation of the DME pseudolite appliqué and its integration with an existing operational DME/TACAN. It also shows the design of the preliminary pseudolite signal that was tested. This paper examines the on-air reception performance of the DME pseudolite synchronization and data signal. It compares the results with model estimates to understand the effects of flight on actual reception error and to aid with practical forward error correction (FEC) design.
2 DME AND DME PSEUDOLITE
The next generation of airspace is being developed with GNSS taking a leading role in navigation. With GNSS becoming the primary means of navigation for many operations and a key enabler of automatic dependent surveillance, its susceptibility to interference is a growing concern. Hence, terrestrial navigation aids, and especially DME, will still have a major role to play in supporting future airspace operations. These systems may provide navigation should GNSS be unavailable or its use unadvisable. As such, DME should provide operational capabilities similar to many of those gained from GNSS.
DME/TACAN has long been the backbone of the aviation navigation infrastructure around the world. In the conterminous United States (CONUS), there are over 900 high-power DME (1 KiloWatt or KW) and TACAN (3.5 KW) stations serving the en route and terminal air traffic control domains. In addition, there are other low-power (100 W) DMEs that are used for approach and landing only. The DME and TACAN system locations used in CONUS are shown in Figure 1. Traditionally DME was used to find the range to the DME station associated with a Very High Frequency (VHF) Omnidirectional Range (VOR), allowing aircraft to fly from point-to-point (i.e., one station to another). Today, many commercial aircrafts use scanning DME avionics capable of using multiple DMEs (DME/DME) for position solutions. Currently, DME/DME, when coupled with an inertial reference unit (IRU) to bridge gaps in DME coverage, can support area navigation procedures requiring 1.0 nautical mile accuracy (RNAV 1.0). The FAA’s current build-out of the DME infrastructure will eliminate coverage gaps at Flight Level (FL) 180 and above, allowing use of DME/DME for en route navigation in CONUS without the need for an IRU. Additions to the DME infrastructure are also being examined to provide a near-term APNT capability. Given the vast existing investments in DME and its worldwide adoption as an International Civil Aviation Organization (ICAO) radionavigation aid, it seems prudent to further build upon DME to provide capabilities to support future airspaces. To gain these additional capabilities with DME, the FAA and other groups have examined enhanced DME (eDME) concepts (Ertan & Psiaki, 2016; Kim, Kim, Song, Kim, & Kee, 2013; Kim et al., 2015).
DME Infrastructure in the Conterminous United States (CONUS)
2.1 DME operations
DME provides true (slant) range distance to the DME ground station (also called a transponder). To make this calculation, an aircraft interrogator initiates by transmitting pulse pairs on the interrogation frequency of the targeted DME transponder. Ideally, these pulse pairs are Gaussian pulses with a half voltage pulse width (time between the rising and falling 3 decibel (dB) voltage points of the pulse) of 3.5 microseconds (μs). The transponder, upon the reception of the first pulse of the pair, determines it was a valid interrogation by finding a second pulse at an appropriate time delay relative to the first pulse. After determining that it has received a valid interrogation, the DME transponder opens its dead time gate and becomes nonresponsive to other interrogations. This allows the transponder to be dedicated to making the reply. The reply is transmitted a fixed period, the “reply delay,” after the time of arrival (TOA) of the first pulse of the interrogation. The reply is transmitted on the reply frequency, which is offset from the interrogation frequency by ± 63 MHz. Figure 2 presents the operation of the DME transponder with the top and bottom showing the reception and transmission side of its response, respectively. True range is calculated by the aircraft from the round-trip time from the transmission of the interrogation to reception of the corresponding reply and conversion of the slant range using the aircraft’s barometric altitude. The nonresponsive period typically extends at least 10 μs beyond the transmission of the reply.
DME station response based on X channel reply delay & dead time. Depiction based on Kelly and Cusick (1986)
A transponder may not reply to a given interrogation for two reasons. First, it may be in the process of replying to another interrogation. This process, as discussed above, can cause the transponder to be nonresponsive to interrogations for 70–100 μs. Second, the DME may be sending its Morse code identifier—a sequence of dots and dashes generated by a series of DME pulse pairs transmitted at a rate of 1350 Hz on the reply frequency that provides the three or four letter FAA designation for the facility. The transmission period of a dot and dash are typically 100 and 300 μs, respectively, and there is a gap of 100 and 300 μs between the character of each letter and between each letter, respectively. These Morse code transmissions can last over 6 seconds, are sent every 30 seconds, and have priority over interrogations. During its transmission period, the DME also does not respond to any interrogations.
The DME system has several codes that can be used on the same frequency. These DME codes have different values for the time between the first and second pulse of an interrogation and reply. Different codes also have different reply delays. Table 1 shows the values for the DME X and Y codes.
Parameters for interrogation and reply for DME/N codes
The DME operation has several limitations (and opportunities). The current specifications allow for deviations in the reply delay and transmitted signals resulting in its specified range accuracy of approximately 0.2 nautical miles (nm). Additionally, inherent in the two-way interaction is a capacity limit as the transponder can only reply to a finite number of interrogations—a DME transponder typically is limited to transmitting 2700 reply pulse pairs per second (ppps) ±90 ppps. While some newer transponder can emit more, emitting more effectively reduces the reply efficiency—the percentage of interrogations that receive a reply (Lo & Enge, 2012). Finally, as DME signals do not provide data, with the exception of the Morse code identifier, there is no method of providing integrity warnings other than shutting down the DME operation.
2.2 Enhanced DME
The Enhanced DME (eDME) design provides capabilities to address the limitations of legacy DMEs. Most importantly, it adds pseudo-ranging and data to legacy DME transmissions without impacting DME operations. This enables DMEs to act as pseudolites, using the same frequencies and pulse pairs used in legacy DME operations. Pseudo-ranging capability improves both capacity and accuracy (Li & Pelgrum, 2013; Lo, Enge, & Narins, 2015). The FAA APNT efforts had developed two mutually compatible concepts for DME pseudolites. The first is a “non-priority” pseudolite based on pulse pair position modulation (PPPM); the second is a “priority” pseudolite based on carrier phase modulation. Both forms offer an evolutionary path to transition from traditional DME to a highly capable eDME. While the priority pseudolite provides greater capabilities, it requires new or modified DME transponders (Li & Pelgrum, 2013). The non-priority, or PPPM pseudolite, however, can operate on an existing transponder without modifying the internals of the transponder (Lo et al., 2015) and is the topic of this paper.
2.3 DME PPPM pseudolite
The concept of a DME PPPM pseudolite is to use the basic functionality of a DME transponder to generate a pseudolite signal. A DME transponder sends replies in response to interrogations from an aircraft interrogator. Imagine a static interrogator that sends a DME transponder a precise, pseudo-random sequence of interrogation pulse pairs. The interrogations would cause the DME transponder to transmit that same sequence with reply pulse pairs that can be used to provide a ranging and data signal. To the transponder, the static interrogator/DME PPPM generator is just like another aircraft. The DME PPPM pseudolite generator is essentially this static interrogator. Figure 3 illustrates how DME PPPM uses the basic DME functionality with a static interrogator appliqué. As the appliqué can exist external to the DME transponder, it is compatible with any existing DME transponders. It is also interoperable within today’s DME system as it behaves like an aircraft interrogator. It is also treated as non-priority because its interrogation is just like any other aircraft and does not have priority over other aircraft interrogations.
DME (left) and DME PPPM operations (right)
A preliminary design of the pseudolite signal was developed and implemented to provide synchronization, ranging, and data, as well as to account for losses due to having to operate with other aircraft interrogations (Lo & Enge, 2012). The signal structure of this design uses 500 ppps. First, a second is equally divided into 20 blocks of 50 milliseconds (ms) each. Then each block contains 25 equal segments of 2 ms, within which a single DME pseudolite pulse pair can be transmitted at a set time relative to the start of the segment. This is shown in Figure 4. For synchronization, the times are known a priori. For data, there are 2N possible transmit times to represent an N bit symbol.
Basic structure of preliminary DME PPPM design
As these pulse pairs are used to interrogate the DME transponder, some losses can be expected due to interference from other DME interrogators. The design anticipates losses and incorporates the use of forward error correction (FEC) to compensate for the effect of losses on data. The preliminary design uses Reed Solomon FEC to correct for data erasure and errors.
Test transmissions were created for the on-air tests discussed in this paper. Of the twenty 50 ms blocks in each second, six blocks are used for synchronization and 14 blocks are used for data, as shown in Figure 5. The figure indicates each synchronization (“sync”) block in a given second with the different numbers as they may be different. In some tests, the three sync blocks from the first half second were repeated in the latter half, simplifying the search. Generally, the sync blocks are repeated every second. The data transmission uses a unique data sequence. The resulting DME PPPM sequence, consisting of sync and data, is played through the Universal Software Radio Peripheral (USRP) for 28.6 seconds. DME PPPM transmissions are then stopped for 0.4 seconds to allow our hardware to restart the playback in a manner that minimizes processing delays that would affect the timing of the transmission. Hence, the transmission repeats every 29 seconds to be maximally out of phase (for periods in integer seconds) with the DME Morse code station identification (“IDENT”) sent every 30 seconds.
Sync and data DME PPPM blocks used in March 2015 tests
2.4 DME PPPM pseudolite generator
A prototype DME PPPM pseudolite generator was built to demonstrate and evaluate the on-air performance of the pseudolite signal. The generator consisted of a laptop computer, a USRP, and a precise GNSS steered oscillator. The laptop computer digitally generates a sequence of pulse pairs and provides this to the USRP. In our tests, the laptop uses a previously generated (prerecorded) sequence of pulse pairs to provide the DME PPPM transmissions. The GNSS steered oscillator provides precise timing for the transmission of the pulse pairs. In an operational system, a hardened and interference-resistant timing source should be used. The USRP converts the digitally generated pulses from the laptop to an analog signal that it transmits from its antenna output port. The antenna output port is then connected to the DME/TACAN via a series of protection circuitry. This is shown in Figure 6.
Laboratory test of DME PPPM pseudolite generator with playback computer (left) and receive computer (right)
The DME PPPM pseudolite generator was installed on a modern Moog MM-7000 DME/TACAN transponder. Rather than interrogating over-the-air, the system coupled into the cable between the transponder and its antenna. Protection and isolation components (directional coupler, attenuator, and isolator) were added to prevent high-power DME transmissions from damaging the circuitry of the PPPM generator while allowing low-power PPPM generator signals to interrogate the DME transponder. It also prevented the PPPM generator signals from going to the antenna and being transmitted out. To the DME transponder, the PPPM generator transmissions looked like that from an aircraft. The block diagram of the coupling setup is shown in Figure 7, along with the approximate signal strength of the PPPM and DME transponder reply signal at each node.
Integration of DME PPPM pseudolite generator with existing DME transponder
The block diagram of the DME/TACAN is shown in Figure 8. The figure shows that the PPPM, along with simulated traffic, was injected via the standard radio frequency (RF) port. Ohio University made modifications to the unit, which are described in more detail in Pelgrum et al. (2015). While these modifications were not needed for PPPM, they allow for the transmission of priority eDME and for test measurements. Priority eDME signals that are always transmitted and have priority over other pulse pairs, such as interrogation, require modification. Both of these eDME transmissions were tested simultaneously and can coexist.
Modified Moog MM-7000 TACAN for enhanced DME tests (Pelgrum et al., 2015)
2.5 On-air test
The on-air test, using the MM-7000 with a dB Systems 510A omnidirectional DME antenna, was conducted at Ohio University Airport outside Athens, Ohio. The system was operated on DME channel 83Y, which has interrogation and reply frequencies of 1107 MHz and 1044 MHz, respectively. As it used the designator “XJJ” and was operated experimentally, the only user/interrogator of the transponder would be the Ohio University test aircraft.
Since no other traffic would be using the transponder, interrogations were added to simulate a fully loaded DME with 100 aircraft (International Civil Aviation Organization (ICAO) 2006). Of the 100 aircraft, two operated in the search mode and the rest in tracking mode. An aircraft interrogator in search mode interrogates at a much higher rate than one in tracking mode. The result was an average of 3007 interrogations per second, with a minimum of 2520 and a maximum of 3579, with the simulated sequence repeated every 24.73 seconds. This period was used to minimize overlap with the 29- and 30-second repeat period of the PPPM and IDENT transmissions, respectively. Air traffic was modeled to be flying in and out of DME coverage. This resulted in a varying signal power of between −65 to −90 decibels relative to 1 milliWatt (dBm). Additionally, the MM-7000 transmitted priority eDME, a pseudo-random pulse pair sequence, at 250 ppps to test the carrier phase-based pseudolite. This was triggered in the MM-7000 internally and had priority over other interrogations. This sequence was repeated every 30 seconds. Finally, the MM-7000 self-interrogated 100 ppps for monitoring purposes. When fully loaded, there was effectively 3100 to 4200 non-priority interrogation and 250 priority reply ppps.
On the reception side, data collection equipment was installed both on the ground and in the Ohio University Beechcraft Baron used for the flight test. The raw intermediate frequency (IF) signal was collected on the ground and in the air by USRPs connected to data collection computers. The suite was part of the APNT flight test suite developed by Ohio University, which includes an equipment rack containing data servers, integrated GPS/INS truth system, and precise time. This system was developed and used for several flight tests (Li & Pelgrum, 2013; Pelgrum et al., 2015). Collected signals were time tagged using GNSS and later processed to determine the TOA of the half amplitude point of the first pulse. Post-processing was also conducted to further refine the GNSS-based time tag and, hence, the TOA.
The Beechcraft Baron carried two DME antennas, as seen in Figure 9. One antenna was used for data collection and to feed a commercial DME interrogator (DME 2100). The other antenna fed into a DME signal analyzer (EDS-300).
DME antenna locations under Ohio University Baron aircraft
Data from one ground test and five flight tests conducted in March 2015 are available for evaluating the performance of DME PPPM. The ground test provided data collected at the DME transponder. Table 2 shows the flights and their corresponding altitudes.
March 2015 flight altitudes
3 DME PPPM PROCESSING
Processing DME PPPM takes several steps. First, it is necessary to align the receiver time with the one-second DME PPPM frame using the synchronization transmission. After alignment, the data symbols can then be demodulated. Because of errors and erasures on the data symbols, the data are encoded with FEC to facilitate the decoding of the data. Other steps may also be employed to improve the processing performance. Estimating drift rates can help with data demodulation while conducting a coarse acquisition to refine the TOA search space can help with processing speed. The overall DME PPPM synchronization and data processing flow is shown in Figure 10.
DME PPPM processing flowchart
The first step is alignment using the synchronization sequence. The basic synchronization process is similar to the acquisition/search process in a traditional DME interrogator. Traditional search starts with the interrogator transmitting a series of interrogations over a short period of time with a random time delay between each interrogation. At the same time, it receives the various replies from the transponder that it interrogates. As the receiver knows the time offsets between its interrogations, it looks for a set of reply pulse pairs that have the same time offsets. The search can be accomplished by correlating the interrogation time series with the reply time series as shown in Figure 11. The search should account for error sources such as missing replies and variations in the time of arrival. The variations are due to factors such as ground interrogator clock drift, aircraft clock drift, reply delay errors, etc., as well as aircraft motion. In DME PPPM, the synchronization sequence is used in place of the interrogation sequence, so the aircraft acquires the synchronization frame by matching the received reply time series with the known synchronization time series.
Correlation of interrogation with reply pulse pairs to acquire initial range estimate
The processing employed for our analysis uses correlation to make the match. A segment of data, typically 1 to 2 seconds, is used to perform correlation for the initial alignment. The estimated TOA of each pulse pair is rounded to the nearest 100 nanoseconds (ns). The result is used to create a uniformly sampled Boolean strobe sequence with each sample spaced 100 ns apart that is large enough to encompass the first and last sample of the data segment. The sequence is Boolean with one representing a pulse pair arrival and zero representing anything else. A Boolean sequence using the same spacing (i.e., sampling rate) and based on the ideal sync sequence is also generated—in GNSS terminology, this would be the code replica. Because of the error sources, the correlation must allow for some variation in the times of arrival relative to the ideal sequence. We define acceptance tolerance as the allowed maximum deviation from the nominal time of arrival relative to the synchronization. The Boolean sync sequence/code replica is modified so that samples within the acceptance tolerance of the ideal sync TOA also are set to one.
Once initial acquisition is accomplished, an alignment is performed over a shorter segment of data. In the paper, every half-second is used. For the alignment, the search span is narrowed to a smaller time window centered about the expected start of the sequence given the acquisition or previous alignment. The result of the alignment yields the number of presumed synchronization pulse pairs found, as well as their deviation from the ideal. The alignment can be used to determine range, and error may be used to estimate drift.
Once the frame is established, the data can be demodulated because the allowable times for the PPPM transmissions will then be known. Data are demodulated by examining whether a reply pulse pair is received at an allowed time. While there should be only one PPPM data pulse pair in each segment, it is possible to have no or multiple pulse pairs arrive at an allowed time within a given segment, which results in an erasure. Data symbol errors are also possible.
3.1 Factors affecting PPPM processing
Different processing can affect performance of the DME PPPM. Identification of pulse pair TOA is a key first step. Traditional half amplitude methods can be employed. Other processing can also be used as long as it is consistent for all DME PPPM pulse pairs. That is because the DME PPPM processing relies on the differential TOA between its pulse pairs.
There are correlated errors that build up over time. These errors are due to aircraft velocity, aircraft clock drift, and ground clock drift. The synchronization pulse pairs, which arrive at known time offsets relative to each other, can be used to estimate the cumulative effect and mitigate some of the error. This can aid correct identification in the data pulse pairs.
Acceptance tolerance is another important parameter. It can significantly affect the number of correct sync and data symbols found as well as errors and erasure rates. Acceptance tolerance affects the number of DME PPPM pulse pairs found, as well as reply pulse pairs misidentified. For example, a larger acceptance tolerance will result in finding more DME PPPM pulse pairs. However, the larger acceptance tolerance will also result in more misidentifications, where a non-PPPM DME reply is identified as a DME PPPM transmission.
For synchronization, having a larger acceptance tolerance leads to having more correctly identified pulse pairs, which can lower pseudorange error as it increases the number of pulses averaged. However, misidentifications can increase pseudorange error and can cause errors in the drift estimate, which can then affect data. If the acceptance tolerance is small, the pseudorange error caused by a misidentification is small. For data, having a larger acceptance tolerance will lead to identifying more data pulse pairs, but it will also lead to more errors and erasures. In both cases, we want to choose an acceptance tolerance where the gains from getting more correct DME PPPM pulse pairs outstrip the cost of errors or erasures. So while acceptance tolerance affects both sync and data, the optimal tradeoff between finding more DME PPPM and having more misidentification may be different. Hence, it may be beneficial to use different values of acceptance tolerances for sync and for data.
4 SYNCHRONIZATION PERFORMANCE
We first examine the performance of the DME PPPM synchronization transmission. Synchronization in DME PPPM allows the user to establish the DME PPPM frame and determine the pseudorange. From the sync pulse pairs, the time of transmission indicated by the transponder is established. That along with the measured time of arrival yields pseudorange. In this section, the amount of synchronization pulse pairs received and the variation in their time of arrival is examined. The number of pulse pairs found drives ranging accuracy. The ranging accuracy of DME pulse pairs and the effects of averaging are discussed in other papers (e.g., Pelgrum & Li, 2015).
4.1 Synchronization rate
The ability to acquire and align was tested on data collected on the ground and in the air. We examined the reception rate, which we define in this paper as the number of DME PPPM pulse pairs received divided by the total number of DME PPPM interrogations to the transponder. Not all of these interrogations will result in a reply, so we will have missing DME PPPM pulse pairs just from interacting with the transponder. Examination of synchronization pulse pair reception from both on-air data and modeling indicates that the missing sync pulse pairs are mostly due to them not being transmitted as other interrogations and transmission have preempted them.
The ground test provides the best-case, on-air scenario as the receiver is static and is close to the transponder, thus receiving a higher strength signal than can be typically expected in the flight. Figure 12 shows the percentage of synchronization pulse pairs identified each second for the ground test. The percentage is calculated each second by tallying the total number of identified DME PPPM pulse pairs by the total number of DME PPPM pulse pairs (i.e., 500). It initially starts at about 95%, as there is no simulated traffic loaded at the transponder. Hence, it only competes with the 250 ppps priority eDME and the 100 ppps monitor signals. Afterwards, the simulated traffic load is added. With other traffic competing, we measured an approximately 75% reception rate of sync pulse pairs.
Percentage of synchronization pulse pairs received (static ground test, March 9, 2015)
These results can be compared to analytic and simulation models. The analysis uses the model for reply efficiency, the percentage of interrogations that are replied, developed in Lo and Enge (2012). The analysis models the competing incoming pulse pairs and is modified to account for the priority 250 ppps that do not have to compete. However, although the model is mostly generic, it does account for some unique features of the MM-7000 that improve performance. It is important to note that while the DME is loaded with up to 4200 non-priority ppps, it will tune its receiver sensitivity so that it only responds to 2700 ppps. Thus, interrogations received with very low power are ignored and only higher power interrogations like the DME PPPM that are above the sensitivity level may be replied to. Figure 13 shows the comparison of the mean number of reply pulse pairs and reply efficiency based on an empirical model and 100 simulations using a modified version of the model from Lo et al. (2015). At 2700 total replies per second, including 250 ppps beat signal, the results show that our MM-7000 DME/TACAN should issue replies to around 74% of interrogations to meet its sensitivity level with the losses due to competing interrogations. This matches reasonably with the field test results though the comparison should not be exact as the model is an estimate for several reasons. First, we do not have the exact number of interrogations the transponder is sensitive to. Second, the measured reception has errors due to effects such as blockage and propagation that are not part of our modeling calculations.
Mean reply efficiency (simulated), assuming 250 priority beat pulse pairs per second and Y channel transmission (and 8 usec deadtime)
There are periods in Figure 12 where the reception rate was significantly worse. The regular reduced reception periods are anticipated. Most are due to the DME Morse code ident transmission, which occurs every 30 seconds and lasts for about 5.5 seconds for our transmitter. During the transmission period of each Morse code dot or dash, the transponder will not transmit any replies. Other instances occur during the last second of the 29 second DME PPPM cycle used. During that second, the DME PPPM generator only interrogates for 0.6 seconds to allow for time to restart. While we only attempt to transmit 0.6 seconds of PPPM (300 pulse pairs = 500*0.6), we still process it like a normal second and divide by 500 resulting in the lower percentage. Thus, the reduction during the 29th second is due to our calculation methodology rather than the PPPM reception.
The performance of the synchronization transmissions in flight is similar to that on the ground. Figure 14 shows the percentage of synchronization pulse pairs identified each second for the flight test on the afternoon of March 10th. Excluding the previously discussed IDENT and restart periods, about 70–75% of sync pulse pairs were received and acquired. However, there is more variation with flight data. Additionally, there remain some periods where reception is noticeably poorer than the 75% level.
Percentage of synchronization pulse pairs received (flight test, March 10, 2015, afternoon)
The periods of lower reception are examined by comparing the sync reception rate with reception of all pulse pairs from the ground station. Figure 15 shows the comparison of the sync pulse pair reception rate and total pulses received during a period of poorer reception. The plot shows that the reception levels between the sync and overall pulse pairs match well and that the sync losses correspond to lower reception in general. The source of the lower reception is the poor signal strength during these periods. Figure 16 shows the mean signal strength and the total number of pulse pairs received over a period of poor reception. Again, there is good correlation, indicating that the lower reception rates are caused by lower reception power.
Sync pulse reception rate and total pulse pair reception during period of poor reception
Total pulse pair reception and mean signal strength during period of poor reception
One cause for the lower signal strength is line-of-sight attenuation from aircraft maneuvers, such as banking away from the transponder. Figure 17 shows the sync reception rates plotted along the paths of the various flights. At this level of resolution, the reception rates are generally at the 70–80% level. Figure 18 shows a zoomed-in view of one section to show more detail. This section was specifically chosen as it is one of the poorer sections for reception. The sync reception rates are poorer at turns that bank the antenna, located on the bottom of the aircraft, away from the ground transponder. Other instances of poor reception rates are near the airport, where we are at low altitudes and may have poor line-of-sight due to blockage from local buildings, such as hangars or the airframe. Even with the lower reception rates, the worst case was still not below 50%. This level is adequate for synchronization—especially when using a half second to a full second worth of synchronization pulse pairs (100% would be 75 and 150 pulse pairs, respectively). Current DME avionics use about 30 ppps for acquiring a transponder.
Synchronization pulse pair reception rate for all flights
Synchronization pulse pair reception rate for all flights (zoomed)
The statistics for sync reception rates for the ground and flight tests are provided in Table 3. These statistics exclude IDENT and restart periods. The table also shows the cruise altitudes of the flights. The worst performance occurs for flights at lower altitudes, which is consistent with the results of the previous discussion. Again, this suggests that at low altitudes there are some reception issues perhaps due to obstructions or multipath (Pelgrum & Li, 2015).
Mean and standard deviation synchronization reception rate for each test conducted
4.2 Synchronization error
There are two possible mistakes that can be made with synchronization. One mistake is an erasure, whereby the synchronization pulse pair is not transmitted (or not received). The non-transmission is usually due to the sync transmission being superseded by an earlier transponder transmission, such as a reply to an aircraft. As seen previously in Figure 13, for the level of traffic experienced by our transponder, such an event occurs for about 25% of the sync pulse pairs. This event is expected and a standard part of DME operations. A potentially more problematic event is an error whereby a reply to an aircraft is misidentified as a synchronization pulse pair. This can occur if there is a reply that is transmitted slightly before a sync pulse pair transmission time. In this case, the receiver would accept the reply as a sync pulse pair if it is within the acceptance tolerance resulting in a faulty measurement. Where this is potentially problematic is if the synchronization pulse pairs are used to estimate the TOA measurement drift. A faulty measurement can bias the drift estimate. Fortunately, this scenario should not occur too often and significant deviations may be detected and excluded. Additionally, the deviation is already limited to the level of the acceptance tolerance.
Figure 19 shows the variation of the identified sync versus an ideal TOA for the ground test using the first sync measured as a reference zero error. Hence, it essentially measures the relative drift of the aircraft and ground clock as well as aircraft motion. Not surprisingly, variation of sync from their ideal TOA in the air was found to be larger due to the greater motion in the air during any given time period. We also can see sync errors within the same second having some jumps. These outliers are potentially from misidentification or poor measurement of TOA.
Measured sync TOA variation relative to ideal (at different times, in seconds, from start – see legend), March 9, ground test
5 DATA DEMODULATION PERFORMANCE
There are more forms of loss with PPPM data than with synchronization. The main erasure and error types associated with PPPM data are shown in Figure 20. There are several ways to have data erasure. An erasure can occur if the receiver cannot identify a DME PPPM data pulse pair or if it identifies multiple DME PPPM data pulse pairs in a segment. The former case occurs if the PPPM data pulse pair is not transmitted because an earlier transmission superseded it. This is only an erasure if there is not another pulse pair within the acceptance tolerance of an acceptable PPPM data transmission time. This is the Type 1 erasure shown in the figure. The latter case occurs because a receiver may not be able to decide on a correct symbol if it finds two or more acceptable pulse pairs in a segment, so it declares an erasure. One way this occurs is if the PPPM data pulse pair is transmitted, but there are also one or more reply pulse pairs that arrive within the acceptance tolerance of another PPPM data time(s). This is a Type 2 erasure, as shown in the figure. This type of loss would not occur with synchronization. Another way would be if the PPPM data pulse pair is not transmitted, but reply pulse pairs arrive within the acceptance tolerance of two or more PPPM data times (Type 3). An error occurs if a PPPM data pulse pair is not transmitted and there is one reply pulse pair that arrives within the acceptance tolerance of a PPPM data time. With synchronization, this would be an erasure and not an error.
Types of error and erasures in DME PPPM data (when there is overlap, only the early pulse pair is transmitted)
5.1 Data reception
As indicated previously, it is expected that data reception rates would be close to, but not as good as sync, as there are more ways to miss data pulse pairs. The actual performance depends on factors such as processing and acceptance tolerance. In the assessments shown, an acceptance tolerance of 600 ns is used as the baseline. Figure 21 shows the comparison of the data and sync reception rates from the ground test. As the true data symbols are known, the number of correct data symbols are calculated and shown. As expected, there is a strong relationship between the performance of sync and data with the sync performance being slightly better.
Percentage of correct data (top) and synchronization pulse pairs received (bottom) (static ground test, March 9, 2015)
The relationship between the two rates is shown in Figure 22, which plots the sync against the correct data symbol reception rate for the same time period. From the figure, one can see that there is a linear relationship, with a slope slightly less than one indicating that the data performance is slightly worse than sync.
Correct sync symbol vs correct data symbol reception rates
The overall correct data reception rate for some of the tests is presented in Table 4. Compared to the sync performance shown in Table 3, the data reception is around 8–14 percent lower. The reduced reception rate from sync to data is worse in flight at about 12 to 14% lower. Aircraft motion and banking can reduce reception rates.
Correct data reception rate statistics (600 ns acceptance tolerance)
We can see the effect of acceptance tolerance by examining the data reception rate at two different values: 1000 ns and 600 ns. Table 5 shows the data reception performance for an acceptance tolerance of 1000 ns. Compared with Table 4, 600 ns has better reception performance by 5.5-6.5%. This is because the reduced misidentification outweighs the reduced detection. The error and erasure rates are larger than calculated from modeling. This is understandable, as the models do not account for variations in TOA that are experienced.
Correct data reception rate statistics (1000 ns acceptance tolerance)
5.2 Data Erasure and Error Rate
The erasure rate is presented in Table 6. Due mostly to propagation and aircraft blockage effects in the air, results from flight have a larger mean erasure rate than from the ground. We compare this result to the probability of symbol erasure calculated from analysis and simulation. The analysis and simulation use the models developed in Lo and Enge (2012) to model interaction at the transponder. The analytical model is statistical and more conservative while the simulation models the interaction. These models do not account for all errors, such as reception effects due to the signal-to-noise ratio (SNR) and propagation effects. Model erasure rates for acceptance tolerances of 600 and 1000 ns are shown in Figure 23. The nominal operating point is at approximately 3300 interrogation pulse pairs per second. This is calculated from taking 2700-250 beat pulse pairs of replies and dividing by our previously calculated reply efficiency of 74%. The erasure rates from our testing are higher than from the models (about 30% and 41% for the analytic, 36% and 40% for simulation for 600 ns and 1000 ns acceptance tolerances, respectively). The ground test has about the expected erasure rate from the models, perhaps due to having less propagation and blockage effects which are not modeled.
Erasure rate statistics (600 and 1000 ns)
Symbol erasure rates from analytical (dashed) and simulation (solid) model for two acceptance tolerances (600 and 1000 ns)
The error rate is presented in Table 7. This can be compared to the results in Figure 24, which shows the probability of symbol error calculated from analysis and simulation (Lo & Enge, 2012). Results using an acceptance tolerance of 600 and 1000 ns are shown with the analytic model having error rates of about 1.5% and 2.6%, respectively. Simulation results are a little higher at 2.4% and 3.8%, respectively. As with the erasure rate, error rate is higher than expected from the analytic model. The difference is even greater with error. The analytic model results help suggest why an acceptance tolerance of 600 ns performs better as it has lower erasure and error rates about 3 and 1.5%, respectively.
Error rate statistics (600 and 1000 ns)
Symbol error rates from analytical (dashed) and simulation (solid) model for two acceptance tolerances (600 and 1000 ns)
The flight test and demonstration illustrated an important point. If we go by model and ground results only, using 50% of the symbols for FEC should be adequate. From our experiments, we saw that this level of FEC was adequate for the ground, but not for the air. Indeed, the flight results demonstrated higher levels of error and erasures than from the models due to reception effects that are not incorporated. However, this is not a problem—even using 75% of symbols, FEC should result in over 500 bits per second (bps) of data. This should be enough to provide regular updates of station information and integrity, as well as additional services and authentication (Lo et al., 2015).
6 CONCLUSIONS
The paper presents a field-tested implementation of enhanced DME pseudolite using pulse pair position modulation. These tests show that PPPM pseudolite can bring eDME to existing DME without needing to modify the transponder hardware. A DME PPPM generator appliqué was created and integrated into a commercial DME transponder. On-air testing demonstrated the performance of DME PPPM.
The performance of the DME PPPM synchronization and data transmission is assessed for ground and flight tests. The paper demonstrates that synchronization performance and availability is generally good and should be at least as good as the nominal availability of DME. Data reception is a little worse than synchronization. However, losses are anticipated and properly implemented FEC could result in high-message demodulation probability with a reasonable data rate (∼500 bps). Furthermore, some improvements may be possible as its performance is worse than previously developed model results. Overall, the results are good—especially since the performance was achieved under the presence of injected traffic representing a fully loaded DME, a condition that is currently rarely encountered (Alder, Thomas, Hawes, & DiBenedetto, 2008).
Additional analysis on the data performance will still be conducted. This paper provided some basic error and erasure statistics. Detailed analysis will help with design and selection of optimal parameters for the design. While the FAA is not currently pursuing enhanced DME for APNT, the FAA and the US Department of Transportation is currently looking at these improved DME concepts to support robust time distribution. Further testing and analysis of DME PPPL is planned for 2020.
HOW TO CITE THIS ARTICLE
Lo S, Chen YH, Enge P, et al. Flight test of a pseudo-ranging signal compatible with existing distance measuring equipment (DME) ground stations. NAVIGATION. 2020;67:567–581. https://doi.org/10.1002/navi.376
DISCLAIMER
The views expressed herein are those of the authors and are not to be construed as official or reflecting the views of the Federal Aviation Administration or Department of Transportation.
ACKNOWLEDGMENTS
The authors would like to thank the FAA Navigation Services Directorate for supporting this work and the members of the FAA APNT Team. We would also like to thank Jaime Edwards and Adam Naab-Levy for their help in the flight tests.
Footnotes
Funding information
Federal Aviation Administration
- Received October 4, 2019.
- Revision received February 10, 2020.
- © 2020 Institute of Navigation
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Jump to section
Related Articles
Cited By...
- No citing articles found.