Wireless Sensor Networks in Airborne Data Acquisition Applications
There are two principal wireless use cases for Flight Test Instrumentation (FTI) applications. The first is to have a wireless link from a Data Acquisition Unit (DAU) to sensors. For the second case, one or more remote DAUs can be connected via wires to sensors, but connect to the rest of a Data Acquisition System (DAS) by a wireless link. This has, to an extent, been a strategy that has been used in rotorcraft where a DAU resides on a rotor hub and is connected using slip plates to the rest of the system. This still requires a physical link however and a wireless RF link would remove the need for such an oft problematic electro-mechanical solution. The focus of this article is for the first use case – wireless sensors – although many of the same challenges apply to wired DAUs.
Acquiring data from sensors in aerospace applications is almost always achieved by physically connecting a sensor to data acquisition hardware. For systems with a high number of sensors, the installation of wiring can be expensive, time consuming and add significant weight to the aircraft. Wireless data acquisition has been of interest to engineers involved with several applications such as aircraft usage monitoring, space data acquisition, but in particular to Flight Test Instrumentation (FTI) engineers.
FTI engineers typically have limited amounts of time to install a system and often have to change or add sensors during a flight test campaign. Wiring harnesses are heavy and bulky and cost significant effort to install. Late changes in sensor requirements can be especially problematic as wiring harness may need to be removed, taken apart and reinstalled. Wireless sensors would be much easier to install and integrate into an existing system. Additionally, wireless sensors are ideal for hard to reach locations where drilling holes for wiring looms or locating a data acquisition in close proximity may be difficult.
There are however a number of challenges to using wireless sensors which has prevented widespread adoption. At the data transmission level, wired sensors have an exclusive link to the acquisition hardware and there is no problem with contention. In a wireless link, several sensors may contend for access to bandwidth. Network contention in an office environment may be solved with a protocol that uses a Carrier Sense Multiple Access with Collision Avoidance (CSMA/AC) scheme but this mechanism introduces problems in an FTI environment. Chief amongst these is the introduction of packet latency and a reduction in the transmit determinism. Data synchronization is also a critical issue for FTI data and a now widely used standard like the IEEE 1588 Precision Time Protocol does not perform well over a wireless link.
Curtiss-Wright has implemented the LXRS protocol from Lord MicroStrain which counters these challenges as follows
- Data contention
- Time Division Multiplexed Protocol (TDMP) avoids collisions and every node transmits in slots (multiple slots per node possible) allowing for flexible sampling regimes
- Time synchronization
- Nodes synchronize to broadcast beacon (± 32 ms)
- High precision real-time clocks on the nodes minimise drift, which resynchronize periodically
There is also a fundamental challenge to do with RF interference with other equipment on the aircraft and due to the movement of sensors or path interference. This interference can result in a loss of signal and thus lost data. Again, the LXRS protocol has data integrity features to mitigate these effects. There is an acknowledgement mechanism at the packet level which can trigger a resend of data in the next time slot. There is also non-volatile memory on nodes which records sampled data with sampling timestamp. This ensures that in the worst case event of transmission failing, data is still available for post flight analysis.
Another challenge is delivering power to sensors. There is little point having a wireless sensor if it needs to be physically connected to an aircraft’s power supply. Sensors can be powered by batteries but these need maintenance to recharge or replace - there are energy harvesting technologies that can supply power to recharge batteries, but they typically generate only limited amounts of power. Thus, it is important to implement power saving strategies. Investigations into the power consumption of wireless sensors have shown that different modes of operation have different typical power consumption levels:
- Logging and processing of sensed data, 5 mW
- Wireless transmission of sensed data, 45 mW
- Sleeping between data samples, 0.02 mW
The conclusions drawn from this data is that, to minimize power drain, the wireless transmission of data should be kept less frequent. One strategy to achieving this is to send data in bursts, rather than in frequent small transmission. To reduce power further, data should be sampled at as low a rate as is possible. Some data sources lend themselves to low sampling rates better than others. Temperature, for example, is typically sampled at low rates – often single or double digit Hz. A sensor such as an accelerometer could require sample rates in the kHz range which will result in significantly more power required. However, while it may be desirable therefore to limit sensor deployment to mostly low sample rate sensors with few higher sampling sources, the regularity of battery replacement may not be a significant concern. For the sensors used by Curtiss-Wright, the worst case scenario is over a day of battery life in continues use, thus it is unlikely that batteries would require replacing more than once a week and for many sensors, depending on usage, it would be on the order of months. For applications like usage monitoring, it would likely be several months between required battery changes.
The strategies noted in this article have been implemented in a wireless module for the KAM-500 (called the KAM/WSI/104), making it possible to now deploy wireless sensors as part of a wired network, or as a stand-alone wireless system. However, while this brings a practical wireless sensor network into airborne data acquisition applications, it is important to note it is not a panacea for replacing a typical traditional wired system. The amount of data being gathered in FTI applications is increasing and wireless, while possible, is not an optimal solution for high bandwidth sensors or indeed data sources like HD cameras. Currently the KAM/WSI/104 is limited to a bandwidth of 256 Kbps and a maximum sample rate per channel of 512 samples per second. Additionally, wireless solutions cost approximately twice that of a wired system.
Instead a hybrid topology is likely the optimal solution, where wireless sensors can be used to reach difficult locations. It is also an attractive solution to add some extra measurements quickly for different test phases or to quickly meet additional requirements that arise during a program without incurring a large installation penalty. In a single ecosystem, the same software and configuration files can be used – in essence the wireless sensors appear to the user as no different to wired sensors. More details on some of this discussion can be found in the Wireless Data Acquisition in Flight Test Networks conference paper.