Open Access
Add to BibSonomy
Abstract
Bringing an optical network infrastructure into civil aircraft cabins is one of the major technological evolutions of the aeronautic industry. During the last decade, copper networks installed in aircraft cabins have proven to be unable to support new technologies and services throughout the life of the aircraft. With severe constraints and high maintenance costs in the aeronautical field, there is also a significant increase in aircraft weight and hence fuel consumption with the current cabin infrastructure. In order to overcome these problems while offering the best trade-off in terms of heterogeneous service transport, we propose therein various optical network architectures for future civil aircraft cabins combining point to multi-point time division multiplexing transport over passive optical networks with wavelength division multiplexing technology. These architectures offer promising solutions for commercial aircraft with capacities up to about 500 seats, while allowing for a significant decrease in operating costs and maintenance/installation time, thus facilitating the intervention on the aircraft.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Air transport is one of the safest and most popular means of transport in the world. Airplanes strive to move faster, more comfortably, and to provide digital entertainment services. About 4.1 billion people per year used airplanes in 2017, an increase of 7.3% over 2016 according to IATA [1] (before the COVID-19 crisis). Most of the time, passengers spend their flight in a very limited space in the cabin, so that there is a high need to provide them with digital services similar to those available on the ground. With the fast development of communication technologies and the explosion of digital uses in recent years, expectations of passengers have evolved pushing airlines to provide new services. The on-board network infrastructure is then rapidly becoming obsolete due to the significant increase in the transmission data rates. In order to maintain the best flight experience, airlines are updating every 5 to 7 years the cabin’s entertainment infrastructure, which has a considerable impact on the aircraft maintenance cost. Hence, a standardized cabin infrastructure which can both manage entertainment services and specific cabin crew information could be of great interest for airlines, in order to support all the cabin reconfiguration during the life of the aircraft with little difficulty. However, many different types of data from different systems are transported in the cabin, each with a dedicated infrastructure, leading to a complex and costly upgrades when new services need to be added. In this context, the ideal network architecture for the future aircraft cabin must be the same for all types of signals, protocols and data rates. This can be achieved by installing an optical fiber infrastructure [2,3].
For several years now, research centers and aircraft cabin equipment manufacturers have been considering a massive shift towards fiber optic systems, as what appeared in Telecom networks several decades ago. It is already known that optical fibers offer many advantages as compared to current copper links: low clutter, weight reduction [4], power consumption reduction, immunity to electromagnetic disturbances, and a very high link bandwidth. The latter point is of crucial importance as it paves the way for definitive and sustainable cabin network infrastructures over the entire life of the aircraft. In this framework, optical networks and Wavelength Division Multiplexing (WDM) systems can significantly improve the design and performance of aircraft cabin information exchange networks. The NASA program AATT (Advanced Air Transportation Technologies) has focused on the fiber transport of broadband signals for avionic systems [5–9]. The RONIA project (Requirements for Optical Networks In Avionics) has specified the technical requirements for the development of optical networks supporting a wide variety of aircraft systems, and also the need for high-capacity transmission on a range of representative military aircraft platforms [10,11]. In Europe, the European FONDA program (Future Optical Network Distribution for Aerospace) from 2005 to 2007 followed by the European FP7 DAPHNE project (2009–2013) have led to a better definition of the suitable optical technologies and protocols adapted to the aeronautical environment [12–14]. However, the road to their mass adoption is still long, due to the specific constraints of aeronautics and the standardization process for the aircraft systems. Meanwhile, the DAPHNE project has clearly described optimized cabin networks as the most likely to lead to substantial gains in terms of performance and weight. Moreover, since several years, the aeronautic industry is moving to new propulsion energies, like hydrogen, which gives hope for greatly enhanced environmental behavior of aircraft. Hence, the power consumption and the weight of systems are of primary importance, and optical technologies can provide efficient solutions.
Among all possible fiber optic options, the disruptive ones do not come from the use of multimode fiber for point-to-point links, but rather from the integration of single-mode fiber technologies into a genuine optical layer transmission and networking medium. This option makes it possible to create a large variety of innovative optical network architectures, and in particular point-to-multipoint ones based on a Passive Optical Network (PON) whose performance, durability and reliability are already proven [15]. Such networks involve all-optical components and devices that are easily scalable and reconfigurable (e.g., wavelength multiplexers, optical add/drop multiplexers, optical space-switches and couplers). This point is of great importance as the cabin and its passengers can be seen as a small access network in terms of triple play services in the telecom sense. The optical network layer supports heterogeneous signal rates and protocols from a large set of data sources, with the ability to aggregate those signals and protocols onto high-bandwidth links, and can cope with various requirements in terms of link redundancy, latency and security. The cabin optical network solution must allow for future network improvements without the need for rewiring the whole aircraft infrastructure. Ideally, upgrades to new or enhanced systems should only require a change of hardware at the end nodes.
It is in this highly changing context that we propose and compare different optical network architecture scenarios for future civil aircraft cabins, based on our experience in the field of interconnection systems for avionics. The sectioning is organized as follows. Section 2 sets out the multiple protocols used in the aircraft cabin, dealing with the different services and constraints. Section 3 then details the generic optical network solution proposed to manage this diversity with an evolving structure. In section 4, we present the different optical architectural configurations selected for the aircraft cabin. The performance of these solutions are compared in section 5, using a relevant set of evaluation criteria. Section 6 focuses on the quality of the optical transmission links using numerical simulation, and a final conclusion is drawn in section 7.
2. On-board communication systems and aircraft constraints
2.1 Data protocols and buses
The communication infrastructure deployed in the aircraft cabin involves an interconnection of the architectures of the cabin systems, entertainment systems (IFE: In-Flight Entertainment) and external communication systems. These different systems use different types of signals and protocols that are quite specific. The main data buses and protocols deployed in the aircraft cabin are: discrete, CAN (Controller Area Network), Ethernet, RF (Radio Frequency) and RS485. They are set up between the processing centers of the three systems mentioned above and the communications nodes installed in the cabin. Analyzing the data buses and protocols deployed in the cabin also requires a detailed study of all the cabin equipment in communication. In this context, an inventory of the cabin terminal equipment is necessary to dimension the interface requirements and the signals carried in the network. This study has been carried out for two types of aircraft: Single-Aisle (S/A) and Twin-Aisle (T/A), as shown in Table 1.
All above mentioned signals can be found in aircraft cabins but not necessarily at the same time (e.g. a cabin with 10 video projectors will not be equipped with seat screens). These buses and protocols correspond to a set of embedded applications and systems on aircraft platforms, all of which should be supported by the proposed optical network infrastructure. They are deployed with different types of such physical media as coaxial cable, twisted pair and optical fiber. Therefore, an independent distribution network is dedicated to each communication system, leading to a heterogeneous and complex infrastructure. As a consequence, the current cabin infrastructure is not very scalable, difficult to operate and expensive to maintain.
2.2 Constrained ecosystem
The implementation of intra-cabin networks requires considering non-independent specific elements, whose interaction produces the following constraints:
- - Cabin network infrastructure: the infrastructure should be able to carry internal cabin signals from equipment such as on-board entertainment servers, service equipment dedicated to cabin crew and passengers and it should be operational for at least 20 years.
- - Architecture of different systems and multiplexing of signals: the definition of the network architecture is a complex process involving various considerations such as :
- • Service aspects: protocols to be supported, quality of service, system and network management constraints.
- • Topological and physical aspects: cabling, signal trans-mission and multiplexing, link and equipment redundancy.
- • Technological aspects: software and hardware components that must be operated and maintained.
- - Environmental constraints: apply to any active or passive equipment in the cabin. Two representative criteria are particularly restrictive: the operating temperature in the cabin required for equipment qualification (from −40°C to + 85°C) and the maximum frequency of vibrations, ten times more severe in flight than on the ground, with 2000Hz for the aeronautical standard DO-160 [16] versus 200 Hz for the Telecom standard Telcordia GR63 [17]. In our case, the vibration constraints impact optoelectronic active components, whose integration are driven by specific assemblies. Therefore, the vibration issue is not addressed here, and we focus on the temperature aspect, which has the greatest impact on the performance of the components in the cabin network. In this regard, it should be considered that the operating temperature range is specified according to the area of the aircraft. Even in a cabin context, which extends beyond the area where passengers reside, the equipment can be installed in against the external structure of the aircraft (ceiling, floor) which will be subject to extreme temperatures, depending on the geographical area. Even if the equipment does not experience extreme temperatures in its nominal operation, it has to be able to start under extreme thermal conditions.
- - Manufacturing constraints: modern civil aircraft are assembled from prewired sections. Connectivity is therefore an essential point in the building process of the aircraft. The mastery of the optical fiber connections and the homogeneity of the optical budget on all the optical fibers deployed in the aircraft must be validated. The use of a limited number of fibers in the architecture is therefore recommended.
- - Ecological constraints: nowadays, aircraft manufacturers are under strong pressure to reduce fuel consumption. As a consequence, optical technologies such as fibers and low-power lasers help to reduce the weight of the aircraft and the decrease in electricity consumption of the cabin communication network.
3. Resilient optical architecture
The current interconnection architecture in the cabin is the juxtaposition of three systems conveying flows of different nature: External Communication Systems, Cabin Systems and Entertainment Systems to transport different data flows and protocols in order to ensure the multi-services deployed in the cabin. These services are deployed via separate networks, using point-to-point links over copper cables. The schematic in Fig. 1 shows the current network architecture used to interconnect the three subsystems of the cabin.
Table 1. Protocols, Buses and equipment in the aircraft cabin
The main objective of this work is to propose optical network architectures for new-generation cabins of civil aircraft. The benefits are twofold: future proof much higher bandwidth and significantly reduced weight and footprint as compared to current copper solutions. These architectures must obviously take modern wireless transmission technologies and new services into account as well as the growing needs of the aircraft's personnel in terms of video monitoring and piloting assistance outside the aircraft. The network infrastructure is based on the classical “three-tier” model coming from the telecommunications field, namely core, distribution and access. According to this hierarchical model, the three-tier network is represented here by specific functional units namely a HUB, a NDB (Network Distribution Box) and a PDB (Passengers Distribution Box). The main driver for this new network architecture (depicted in Fig. 2) is the mutualization of data distribution via the HUB installed at the head of the network. The HUB processes all types of protocols/data coming from the central system, before it is routed over a unique optical distribution network. The HUB, NDBs and PDBs are optically interconnected in sequence and located at specific areas. This cabling principle considerably reduces the number of links required to interconnect all systems and terminal communication equipment. It also guarantees their redundancy and thus the robustness of the architecture to a link failure. The various data flows circulating in the cabin, emerging from the HUB network head then pass to the ceiling, where they reach the NDBs to link the ceiling devices (cabin system). Then the NDBs connect to the PDBs at the floor, feeding one or more rows of seats.
The three network segments are organized as follows:
- • Core from HUB to NDBs: the network core acts as a transport of aggregate traffic; it interconnects the cabin equipment with the entertainment system servers, the cabin system servers and the external communications servers. So, the HUB processes the data and performs the aggregation or the optical deport according to the type of data bus before being transmitted to the NDBs.
- • Distribution from NDBs to PDBs: the NDB is the communication point between the core network and the access tier. It provides the intelligent functions of switching, transport, access to overhead cabin equipment and the link between the NDBs installed in the cabin crown on the ceiling and the PDBs installed on the floor.
- • Access from PDBs to seats: the access part, based on PDBs installed on the floor serves to connect the multimedia terminal equipment of the network installed in the seats. The PDBs are located close to the seats, in the lowest part of the cabin, in order to reduce the length of the access cabling and thus its complexity and weight.
- • Network protocol transparency: WDM networks natively support a variety of different signals and protocols.
- • High capacity scalable network: WDM networks can support a large number of high capacity channels that can be added as traffic grows without any heavy modification.
- • Flexible, secure and reconfigurable: WDM reduces the complexity of the physical layer connectivity, by reducing or eliminating cable overlays, whereas TDM can dynamically change the bandwidth of the allocated wavelength to ensure high flexibility and efficiency in signal transmission. In addition, network flexibility can still be carried out via the use of WDM functions such as Reconfigurable Optical Add-Drop Multiplexers (ROADM).
4. Optical network configurations
Following the approach of the previous section, we will now present and describe some optical network architectures suitable for civil aircraft cabins. These proposals should respect all the constraints related to the geometry and topology of the cabin, the cables and equipment deployment methods, the more stringent environmental specifications than those generally encountered in telecommunications and also the different types of data transmitted in these networks. In accordance with the principle of optical transport of data buses, the network infrastructures proposed throughout our work must guarantee the conversion of low-rate digital RS and Discrete signals to Ethernet, and the optical transport of analog RF and CAN services. As a consequence, each HUB and NDB always include these conversion modules since all these protocols are in-cabin native-based equipment. Moreover, the work on the architectures proposed is motivated by providing new infrastructure services for airlines that could be of great interest like: (1) an easy network management, (2) an easier cabin configuration and (3) a faster architecture installation and/or maintenance. These new services could be a game changer in order to propose an architecture compatible with any aircraft type, whether it is the manufacturer or the number of seats. In this framework, four configurations of optical network architectures adapted to civil aircraft cabins based on a combination of WDM and PON technologies are described in the following subsections.
4.1 PON architecture with CAN/RF/NAS point-to-point connection
The first proposed network architecture (referred to as option A) is based on point-to-multipoint passive optical networks (PON) for Ethernet transport. This architecture uses optical power splitting elements (couplers) to connect the NDB to the multiple PDBs. Being passive, these devices are not electrically supplied nor contain any electronics, which is a very strong guarantee of reliability. The point-to-point part is dedicated to the transport of the CAN/RF/NAS, each on a dedicated optical fiber in order to separate high rate data buses (Ethernet Bus) from deterministic control and command buses at lower rates (e.g., CAN). The choice is made here to transport the data flows to the NDBs with different PONs and therefore using different fibers. Figure 3 shows the functional structure and interface distribution for the HUB, NDB and PDBs with this architecture. The connection of a NDB to the HUB then requires a single bi-directional optical fiber (without considering the protection links) and each NDB is linked to two PDBs, which are themselves connected to the seats (each PDB is connected to 12 seats). The choice of the number of PDBs per NDB and the number of seats per PDB was made after a study of their impact on the complexities of the equipment and the cabling to be implemented in the limited space available in the floor of the cabin.
The Optical Line Terminal (OLT) interface of the PON on the network side is placed at the HUB while the client Optical Network Unit (ONU) interfaces are placed at the NDBs and the PDBs. One ONU at the NDB for the cabin ceiling equipment and one ONU per PDB to connect the seats. It should be noted that an ONU integrates an Ethernet traffic concentrator allowing the multiplexing of flows on the OLT-ONU link. Their bit-rate characteristics are calculated according to the needs of the different equipment connected and the traffic segregation options.
4.2 PON + WDM-dual-fiber architecture
The next proposal (option B) aims at minimizing the number of fibers and connection interfaces deployed using Wavelength Division Multiplexing. This architecture is divided into two different sub-networks. Like option A, a point-to-multipoint part transports the data flows to the NDBs with different PONs using dedicated optical fibers. Then, a WDM link is used to transport RF/CAN/NAS on a single optical fiber, but with a specific optical channel (wavelength) for each service. Figure 4 shows the functional structure and interface distribution for the HUB, NDB and PDBs for this architecture.
Fig. 4. Option B: Block diagram for the PON + WDM-Dual-Fiber architecture including HUB, NDB and PDB.
The OLT interface (i.e. the network side) of the PON as well as two optical MUX/DEMUX (4 channels) are located in the HUB. The ONU interfaces are placed at the NDB and PDB. One ROADM is used at each NDB to direct the traffic coming from the HUB of the multiplexing part, whereas an optical coupler separates or collects the optical signal of the PON. The traffic moves to the NDB in which a ROADM manages three add/drop wavelength channels: one for the NAS, the second for the CAN and the third for RF. Similarly, to the PON part, one fiber per NDB passes through a coupler to divide/group the flows in three parts: one ONU at the NDB for the cabin ceiling equipment and one ONU per PDB to connect a set of 12 seats.
4.3 PON/WDM dual-fiber architecture
The third option (option C) aims at multiplexing all PON signals on the same fiber in order to reduce further the overall number of fibers used. This scheme is based on the point-to-multipoint optical architecture of passive PON networks with WDM, using one optical fiber for each direction of transmission. Thus, the choice is made here to transport the flows to the different NDBs using different PONs, but with the same optical fiber. Figure 5 shows the functional structure and interface distribution for the HUB, NDB and PDBs with this architecture.
The OLT interface of the PON as well as two MUX/DEMUX are placed in the HUB. Eight wavelengths pass through one MUX to regroup them into a single output fiber for each side of the cabin and for a one direction of transmission. The ONU interfaces are placed at the NDB and PDB. Each NDB houses a ROADM and two couplers to route the traffic coming from the HUB.
At the NDB level, a ROADM is used to direct four optical channels coming from the HUB to the local equipment concerned:
- - One channel wavelength dedicated to each PON (λ1 in Fig. 5) is fed to a passive coupler to feed three ONUs. Two of them are placed in the two local PDBs (one PDB connecting 12 seats), and the third one is placed at the NDB to connect cabin overhead equipment.
- - Three channel wavelengths (λ6, λ7 and λ8 in Fig. 5) are used for NAS, CAN and RF transport respectively.
4.4 WDM architecture
This architecture (option D) aims at spatially separating the optical infrastructure into two distinct networks: ODN (Overhead Distribution Network) and SDN (Seat Distribution Network) on two different physical media to facilitate the management of the various signals circulating in the cabin. This architecture is based on WDM with two optical fibers for each direction of transmission. Figure 6 shows the corresponding functional structure and interface distribution for the HUB, NDB and PDB. With this option, four switches and four 8-channel MUX/DEMUX are placed in the HUB. Two ROADMs for each NDB direct the traffic coming from the HUB and an Ethernet switch shares or recomposes the signals from or to the two PDBs. The choice is made here to transport the flows to the different NDB destination on two different fibers. One optical fiber transports the flows to the ceiling equipment, CAN and RF. The other fiber is dedicated to transport the data to the PDBs and the NAS part (cameras and video projectors).
In the next section, we move on to a comparative study of the performance of the different configurations proposed while using a set of criteria.
5. Network parameters analysis
In this part, a comparative study of the different options is performed on the basis of dimensioning parameters involving the topology and complexity of the architectures. The main criteria for comparing these architectures are listed in the following.
5.1 Number of fibers
The number of fibers used per architecture is an important criterion in the aeronautical industry to determine the type, diameter and weight of cables, as well as a vision of the complexity of maintenance and installation of the infrastructure. The counting of the resources involved can be done by exploiting the architectural diagrams presented in section 4. The results shown in Fig. 7 confirm that the hybrid PON/WDM architecture uses the least amount of fiber, while conversely, the pure PON architecture is the one that consumes the most. Then, an estimation was made on the length of fiber based on a set of criteria related to the cabin structure, cable route and the dimension of a single-aisle cabin of 240 seats with a length of 30 meters, width of 6 meters and height of 3 meters chosen for our study. Figure 7 shows the same classification if we calculate the total length of fiber used per architecture, making the hybrid PON/WDM architecture the best candidate to reduce the optical cable size and footprint.
5.2 Number of components
The optical components used in our solutions fall into two categories: passive components such as couplers, MUX/DEMUX and connectors that generate power losses between transmitters and receivers on one side, and active components such as transceivers, Ethernet switches and ROADMs on the other side. These latter components are the most complex and fragile and require ultra-high reliability to maintain performance in harsh operating environments. Therefore, the total number of components used by category is an important criterion for the choice of the optimal architecture, to estimate the electricity consumption of the active components and the overall weight of the communication network. In addition, the number of components has an impact on both the cost and reliability of the process. Using this criterion, we can determine the links power losses and estimate the level of risk of failure of the system. The chart in Fig. 8 shows the variation of the number components used for the different architectures. Overall, the WDM architecture is more consuming in terms of components than the other options. The PON, PON + WDM and PON/WDM architectures require the same number of active components (Transceivers and Ethernet switches) but with varying number of passive components. The equipment and links of these networks are likely to suffer from failures that could affect their proper operation. Therefore, the architecture that uses the uses the largest number of components cascaded components has the greatest risk of failure. This quantitative study allows us to favor the PON + WDM and PON architectures with the least number of components used and therefore lower weight, less consumption, footprint and risk of system failure.
5.3 Link power budget
The performance of any communication link depends of the quality of the equipment used. The link power budget is a way to estimate the performance of the link and its ability to be operational. Table 2 summarizes the distribution of optical power losses between the HUB and different NDBs by architecture, according to the nature of the signals transported, from the shortest (NDB1) to the longest link (NDB5). As expected, the PON and PON + WDM architectures show the same power losses for data transport: 8.4 dB for any traffic point from NDB1 to NDB5. But buses and protocols (RF, CAN…) are transported in two different ways : for the PON + WDM architecture (option B), all protocols and buses are multiplexed on the same optical fiber which then requires the use of passive components in order to properly manage the traffic (the use of MUX/DEMUX and ROADM generates additional loss). For the PON architecture (option A), all protocols and buses are transported on different optical fibers (point-to-point), which means that power split-ting components are not used, hence causing less losses. The PON/WDM architecture transport all protocols, buses and data on the same optical fiber using an optical infrastructure that is adapted to manage the traffic in specific areas of the cabin (use of MUX/DEMUX, ROADM and couplers which generate more loss). For the WDM architecture, the optical infrastructure is composed of two separate networks (ODN and SDN) that run on a specific optical fiber and use only ROADMs at the NDB to manage the traffic in the cabin.
Table 2. Distribution of power losses according to the architecture (in dB)
The results presented in Table 2 confirm that the power loss of the different architectures depends on the number of optical components through which the signal passes between the transmission and receiving stages. As our application addresses a short distance network (a few tens of meters), the loss generated by the fiber is neglected. The use of the PON is strongly motivated here as a highly flexible solution for cabin upgrades of configurations without tearing up the cabling, while keeping the existing network infrastructure unchanged. The loss values presented in the table are compatible with the optical budget of the standardized PON interfaces (typically 25 dB for SFP transceiver at 10 Gbps after allowance for adequate margin) which shows that these optical links can be achieved. Furthermore, the use of optical amplifiers and SOAs in particular is not allowed in an aeronautical context due to various constraints such as stability, power consumption and cost. Hence, power budgets above 25 dB become complicated to handle and do not guarantee a bit rate upgrade beyond 10 Gbps for operation without any optical amplification. As an example, the next step beyond 10 Gbps in Datacom is no lower than 25 Gbps, and a standard SFP25/28 transceiver module would exhibit a receiver power sensibility raised by about 4 dB with respect to a 10 Gbps SFP+ transceiver, all things being equal. Hence, a bitrate upgrade over an installed high loss optical architecture cannot be realistically carried out without amplification. Thus, architecture partitioning would be the way to cut down the power budget by a substantial amount in order to be compatible with future bit rate upgrades, whether they are obtained by a higher baud rate or a higher spectral efficiency.
5.4 Power consumption
The overall power consumption of the aircraft is distributed in two parts. The first part is the power consumption of the on-board equipment, and the other part is the kerosene consumption. In this paper, we are interested in the distribution of the electrical consumption of the equipment according to the different network architectures proposed. The estimation results are shown in Table 3, on the basis of commercially available specifications.
Table 3. Distribution of electrical power consumption by architecture (in watts)
The analysis of Table 3 confirms that the HUB, NDBs and PDBs of the PON, PON + WDM and PON/WDM architectures consume the same amount of electrical energy, as they use the same active components in the different units of the network. Then, the pure WDM architecture consumes the most energy in the NDB. This difference in consumption at the NDB is due to the use of an Ethernet switch, which has been replaced in the other three architectures by a passive coupler. At the PDB level, we use an ONU with an Ethernet switch in the first three options, which increases consumption compared to pure WDM architecture which only uses Ethernet switches. Since the cabin infrastructure counts 10 NDBs and 20 PDBs, we can estimate the total consumption per architecture. These results show that the equipment power consumption is similar in the proposed A, B and C options. A more precise choice of the optimal architecture need the use of another criterion, which is the weight of the infrastructure, directly impacting the fuel consumption of the aircraft.
5.5 Weight of the infrastructure
Weight is a very important criterion in the aeronautical world. The weight of the network infrastructure comes from all point-to-point links that are unique per system or function. In some cases the cabling is reduced in the aircraft to optimize the time spent on maintenance or retrofit. An actual weight saving is achieved when switching from copper to fiber optic as a transmission media. For example, the IFE system installed in the current cabin uses a minimum of 1 km of Ethernet cable with a weight of 41 kg/km to connect the servers to the PDBs. Switching to fiber cable with a weight of 4.2 kg/km hence represents a substantial saving. In the same way, the replacement of the 80 m long coaxial cables of the RF subsystem reduces the weight from 6.8 kg to a few hundred grams by using optical fibers. In addition, a change of antennas (e.g. from 4G to 5G) would not require a rewiring as it would be the case for coaxial cables. Nevertheless, the four proposed architectures have very similar cabling weights as optical cables are very light. Potential weight savings per architecture are then evaluated considering the nature and number of installed electronics and/or optical equipment. These components can be changed over time while the cabling infrastructure is retained throughout the life of the aircraft. Thus, the difference in weight between the proposed architectures comes mostly from the nature and quantity of installed equipment. The classification of architectures by increasing weight can then be estimated by the number of components of the same nature.
5.6 Operations and maintenance of the network
The last point, which is of prime interest for airlines, is network management for operations and maintenance issues. The challenge is here to be able to change the configuration of a cabin quickly and securely from a single access point. The network must easily take this evolution into account and its management must be simple with a centralized optical node. The objective is also to follow the state of the network in real time, to be able to carry out preventive maintenance and to locate a breakdown directly. The preliminary study of these comparison criteria shows that maintenance will be much easier for architectures with fewer optical connections between fibers. Indeed, when the number of links is reduced, the number of failures on a reference period decreases proportionally, as well as the repair and intervention costs. In addition, architectures with colorized components require additional installation costs as compared to non-colorized architectures, but these costs will be profitable over the lifetime of the aircraft. The criteria of reconfiguration, management and security will be given higher priority by airlines than the additional costs that will be expended only once over the lifetime of the aircraft. The evaluation of the architectures on this operations/maintenance criterion can then be estimated by the total number of fibers and connections used to connect the HUB to the NDBs and the NDBs to the PDBs, with the specific components used in the NDBs and PDBs being pre-assembled in the factory in a dedicated box.
5.7 Summary
The previous network parameters study is summarized in the following Table 4, in the form of a multi-criteria analysis. An overall score for each of the four architectures is calculated in the last row, based on a weighted sum of the scores obtained on each of the six criteria studied. The number of stars in this table varies between 1 and 4 depending on the importance of the criterion in ascending order. The weighting coefficients (W/C) are indicated in the second column, with particular emphasis on energy, weight and maintenance.
Table 4. Multi-Criteria analysis of the proposed optical network architectures
As a first outcome, it allows to eliminate the pure WDM architecture from the optimal choices. The use of a combination of WDM technology and PON architecture can offer higher bandwidth and reach and additional benefits over pure PON in these applications. By using dedicated wavelength channels by each part of the cabin network, WDM mixed with PON is considered as a better choice. Therefore, the two architectures based on PON and WDM are the most interesting to enable future proof networking. These architectures support a modular and generic approach, so that they can be used in several applications without any major changes. In addition, they provide multi-level security for several protocols as a network service. As a consequence, options B and C proposed in section 4 are the two architectures selected for the next step, dealing with the dynamic performance of the transmission channels.
6. Optical simulation results
6.1 Simulation goals
The previous section has given a first insight of the implementation of the proposed solutions, but no qualitative performance assessment in terms of very high-speed transmission has been developed. Indeed, new generation optical networks should be able to support current and future services with bit rates of 10 Gbps and beyond. These bit rates are inaccessible to copper-based current technologies even over short distances within the cabin. The purpose of this section is to evaluate the optical transmission quality of the different proposed network architectures and more specifically the two selected in section 5. Numerical simulation makes it possible to build and test different environments in a realistic way, thanks to the use of physical models of the different optical components used (fibers, lasers, filters, MUX/DMUX, ROADM, photodetectors). Another advantage of numerical simulation is the ability to build complex systems involving a large number of links and components (several hundreds), without having to resort to the implementation of bulky and expensive prototypes. For this purpose, we use the commonly used VPI Transmission Maker digital photonic simulation platform [19].
6.2 Component models
The choice of accurate physical component models for the simulation is a key element to build and test different network links. One of the most critical components is the laser source, that has to be adapted to WDM transmission over single-mode fibers, requiring the use of DFB lasers [18,19]. The cost-effective direct modulation scheme is simulated using the “LaserDFB_DSM” VPI model, which is easily customizable with measurement data made on actual lasers. The bitrate is set at 10 Gbps with an average transmitted power of 1 mW at the reference temperature of 25°C. At the other end of the chain, the receiver model includes an avalanche photodiode (APD) [20] with 0.8A/W responsivity and 5.10−12 A/√Hz thermal noise power spectral density. This module can handle both single-mode and multi-mode optical signals and also be used for the simulation of CWDM systems. It can be described on the basis of a predefined responsivity, avalanche multiplication, dark current and noise. In addition, receiver bandwidth and temperature dependence are taken into account by using an equivalent RC circuit. The simulation time window is adjusted to generate a 2048 pseudo-random OOK bit sequence at 10 Gbps, with a simulation bandwidth of 160 GHz. Throughout this simulation, we consider the 8 optical channels of the CWDM standard (Coarse WDM) with wavelength channels ranging from 1470 nm to 1610 nm, using uncooled lasers that consume less energy and are cheaper than their DWDM (Dense WDM) counterparts [21–23 ]. To implement the simulations, we build the whole structure of each architecture, including all links and components through realistic physical models. Each single component is put in the right place in the cabin, as presented in section 4. Connectors are modeled by ideal attenuators while the wavelength filtering components of the network such as MUX/DEMUX and ROADM are described by VPI subsystems (galaxies) using customizable optical filters, power combiners/splitters and attenuators. Figure 9 shows the general structure of a ROADM installed in-line with a dual-fibre WDM link, while Fig. 10 shows the sketch of the ROADM structure as implemented in VPI. This basic one channel add/drop structure can be easily extended to a multiple channel one in a reconfigurable manner. This classical ROADM design using optical filters in a “broadcast and select” scheme allows us to efficiently implement the specific wavelength templates of CWDM components (e.g., several hundred GHz bandwidth), while providing low losses at low cost. In addition, the parameter definition flexibility enabled by this “split” structure allows an efficient control of the physical parameters variations according to environmental conditions. This criterion is crucial in the aviation field.
Fig. 10. ROADM VPI galaxy with single channel Add/Dropping, using a broadcast-and-select bidirectional architecture.
The automated assembly and reconfiguration of these modules is achieved using scripts developed in Python. The optical filter profile can be specified either analytically from a set of widely used filters, or by an import file of experimental measurements on real filters. We use here the analytical method in a first approach, but to get closer to real conditions and add plausibility to the simulated results, we use third-order super-Gaussian shapes with 400 GHz width (FWHM). We use the VPI “Universal Fiber module” as the fiber model. This comprehensive model include both linear and non-linear effects: modal and chromatic dispersion, stimulated and spontaneous Raman scattering and Kerr non-linearity. In vector mode, it includes PMD and polarization dependence of non-linear effects. The short distance and moderate optical power context of the aircraft cabin allows us to deactivate these irrelevant physical parameters to speed up the simulation.
6.3 First link quality assessment: eye-diagrams
In the present context of transparent optical links implementing the cascading of a potentially large number of components, the examination of the eye diagram at reception represents a first step in testing the quality of a digital link. The eye diagram represents the cumulative result of the received waveform in a unit interval. It shows how the received signal fluctuates at the receiver, depending on distortions caused by various physical effects like power loss, chromatic dispersion, channel crosstalk and thermal or shot noise. Therefore, the performance of the high-speed channel is graphically evaluated on the eye-diagram, through essential characteristic parameters that can be directly observed on the detected current, including the “0” and “1” mean values and variance (denoted respectively by i0, i1, σ02, σ12), providing the eye height/width, jitter and crossing percentage. The simulation of the dual-fiber PON/WDM architecture is presented in Fig. 11, for an OOK modulation format at a bitrate of 10 Gbps.
Fig. 11. Optical signal emitted by OLT1 - Electrical signals respectively received by ONU 1, 8 and 18 of the PON/WDM architecture, at a 25°C temperature.
The eye-diagram simulations are carried out in both transmission directions between the HUB and the PDBs. The eye diagrams show the optical and electrical signals of the downstream transmission (from the HUB to the PDBs) for different receiver locations in the cabin: from the shortest link (ONU1) to the longest link (ONU18) at a temperature of 25°C. The eye diagrams in Fig. 11 are open enough to ensure a good decision for the receiver. This therefore suggests a very low bit error rate, as we will see in the next subsection. In a second step, we raise the case temperature of the transmitter laser by 10 °C and activate the Relative Intensity Noise (RIN) to 130 dB/Hz, according to its evolution model as a function of temperature. We increase the thermal noise power spectral density of the receiver to an RMS noise current value of 15.10−12 A/√Hz. Figure 12 shows the simulation results. As expected, there is a gradual degradation of the eye diagrams (as the links length increase), but they remain open enough to assume that the initial bit sequences will be properly processed by the receiver. This conclusion is of course only valid for the temperature considered (35°C), but it tends to show that: (i) the PON/WDM solution is a plausible architecture in terms of transmission quality and is worth further study and (ii) each of its constituent links within the aircraft can be tested individually by simulation, subjected to possible temperature variations. In a future step, this will permit us to refine the required components technologies and performances during the experimental tests.
Fig. 12. Optical signal emitted by OLT1 - Electrical signals respectively received by ONUs 1, 8 and 18 of the PON/WDM architecture, at a 35°C temperature.
These simulations can now be extended to assess the two selected architectures (PON/WDM and PON + WDM) with temperature values ranging from −40°C to +85°C, to comply with environmental standards in the aviation world. This variation has an impact on the transmitter laser frequency as well as the noise power spectral density. However, in order to perform a more rigorous analysis of the performance of our architectures, it is necessary to move to a measurement criterion that is more faithful than the simple inspection of the eye diagram. We will therefore consider the bit error rate at the reception stage.
6.4 Link performance test: bit-error-rate estimation
To make an objective measurement of the link quality, we will calculate the bit error rate at the reception side, using a dedicated VPI module included in the “BER_OOK_Stoch” module. By combining stochastic and deterministic approaches, this module estimates the error probability from the (i0, i1, σ02, σ12) statistics extracted from the eye diagram. In the present simple case of OOK modulation and assuming equiprobable symbols and Gaussian noise statistics, the BER is extracted from the calculation of the Q factor, with Q = (i1 i0)/(σ1 + σ0). This makes it possible to estimate very low BERs (< 10−6), even with only a few thousand transmitted bits. BER estimation can be performed assuming either Gaussian or Chi-square statistics, depending on the presence, or not, of optical beat noise. Post-detection thermal noise, shot noise and inter-symbol interference (ISI) can also be taken into account within the module. For these BER measurements, it was decided to focus on the relevant temperature range to characterize systems within the aeronautical industry. Thus, the simulations were performed for a temperature ranging between −40°C and 85°C, and we chose to present the stochastic estimation results since it leads to similar results as the deterministic approach. To improve the BER estimation accuracy, the simulation time window is increased to an 8192 pseudo-random OOK bit sequence while maintaining the same bitrate and bandwidth. As each simulation is computationally intensive, we did not perform the tests for each HUB-NDBi connection (1 ≤ i ≤5) of the two architectures, but we focused on the link with the worst-case loss (HUB-NDB5) on which we varied the temperature within the targeted range. Figure 13 shows the results of the BER calculation for each temperature step for lasers and photodiodes, for the HUB-NDB5 link in the cabin.
It can be noticed that the two curves are almost parallel and far from each other in the simulated temperature range of [−40°C to +85°C]. This difference in conditions can be explained by the fact that these two architectures offer the same services, but with two different optical infrastructures. The PON + WDM architecture uses 4 multiplexed channels, while the PON/WDM architecture uses 8 channels and involves more nodes and passive components. This leads to a more complex architecture that presents a higher power attenuation for this simulated link. We then observe a more pronounced degradation of signals for the most constrained architecture, especially for longest links, when larger temperature variations are applied. Indeed, channel wavelengths drifts due to temperature variations may cause potential power losses when channels are getting close to the edges of the optical filters pass band.
A second result can also be identified from these curves: the use of Forward Error Correction (FEC) at the receiver extends the operating temperature range for these two architecture configurations. If FEC is not used, the maximum tolerated BER at 10 Gbps is 10−10 and thus the PON + WDM architecture is viable between −27.5°C and 77.5°C, while the PON/WDM architecture works properly between −5°C and +55°C. In the case where we use the FEC at the receiver, the maximum tolerated BER increases to 10−3, increasing the operating range of the PON + WDM architecture to between −40°C and +85°C, and that of the PON/WDM architecture to the range [−22.5°C, +72.5°C]. The use of the FEC allows us to gain 12.5°C towards the “cold” range and 7.5°C towards the “hot” range for the operating temperature of the PON + WDM architecture. However, for the PON / WDM architecture, the “cold” range was increased by 17.5°C and the “hot” range by 22.5°C. The possibility is therefore of real interest here to improve the robustness of the infrastructure.
As a conclusion, using an optical transmission simulator to build realistic models of the two network architectures selected in Section 5 (PON + WDM and PON/WDM) allows us to validate their proof of concept in the context of 10 Gbps optical links. This is first of all valid in a normal operating environment (T = 25°C), but also even if the network is subjected to temperature disturbances, affecting the physical characteristics of the different components used. It remains now to refine the models and physical parameters of the components. This will allow us to describe the precise variation of their performance as a function of temperature conditions, through experimental measurements on real components in the laboratory, primarily lasers, photodetectors and wavelength filters. This step is currently being studied.
7. Conclusions
In this paper, we have discussed the interest of migrating from copper cabling to optical networks in civil aircraft cabins in order to provide a high capacity, flexible, secure and future-proof infrastructure respecting both environmental requirements and aeronautical standards. We started our study by identifying the different services and protocols used in the aircraft cabin and the various constraints for the development of this network. Then, we presented a panel of general architectural options using WDM and point-to-multipoint systems, which are particularly efficient in the optical domain. We then provided performance comparisons based on static parameters to select the most promising architectures.
As a first step to progress towards future solutions, we have assessed different system configurations in terms of signal quality with numerical simulations of the optical transmission in the pre-selected architectures. The comparative study has shown that none of the architecture schemes is well ahead of the others from a purely technical point of view. On the contrary, each solution has advantages over the others if we adopt a more diverse set of criteria. However, the combination of the technical constraints of the aeronautical industry and the multi-criteria analysis made in section 5 led to the selection of two hybrid architectures (PON/WDM and PON + WDM). These two choices are characterized by an optical attenuation compatible with the power budget of standard PON interfaces (about 25 dB) and exhibit comparable power consumption (684.5 W). Although presenting different optical schemes, the two configurations can provide the same network services. Simulations have shown similar performance under normal operating conditions with good signal quality at reception for both architectures. Under more severe thermal conditions, the PON + WDM architecture continues to operate over the temperature range [−40°C to +85°C] while the PON/WDM architecture is limited to an operating temperature range of [−22.5°C to +72.5°C]. These results open the door to future experiments on these two architectures, with tests of commercial telecom components to validate fragments of the network. As such, the physical integration and characterization of a ROADM that can be implemented with any type of architecture will be a key-point. More generally, this work will be based on further optimization studies that can lead to hybrid configurations of optimal architectures for cabin network design for any aircraft type.
While this initial study has highlighted optical architectures with moderate power budgets, further work needs to be done to consider more flexible architectures, but requiring power budgets above 25 dB. This case is not easy to handle and does not guarantee the increase of the bitrate beyond 10 Gbps without any regeneration or optical amplification. Therefore, improvements and optimization studies for these architecture configurations will be carried out to significantly reduce the power budget and thus achieve higher rates. In this context, higher spectral efficiency modulation formats will be also evaluated. Large aircraft cabins with a high number of connection points are likely to support a higher order of spatial and spectral network partitioning. The pathway to increasing the bitrate and capacity still needs to be clarified to ensure true long-term sustainability of optical infrastructures for aircraft cabins.
Funding
Association Nationale de la Recherche et de la Technologie (2019/0118).
Disclosures
The authors declare no conflicts of interest.
Data Availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. The International Air Transport Association, “Traveler Numbers Reach New Heights”, IATA World Air Transport Statistics Released, https://www.iata.org/en/pressroom/pr/2018-09-06-01/, (IATA, 2018), pr. 51.
2. D. E. Anderson and M. W. Beranek, “777 Optical LAN Technology Review,” in 1998 Proceedings. 48th Electronic Components and Technology Conference (Cat. No. 98CH36206), (IEEE, 1998), pp. 386–390.
3. D. Martinec, “Challenges to the Air Transport Industry for Implementing Fiber Optics,” in IEEE Conference Avionics Fiber-Optics and Photonics, 2006, (IEEE, 2006), pp. 1–2.
4. S. F. Habiby and R. Vaidyanathan, “WDM optical backbone networks in aircraft applications: Networking challenges and standards progress,” in MILCOM 2009-2009 IEEE Military Communications Conference, (IEEE, 2009), pp. 1–6.
5. M. F. Alam, M. Atiquzzaman, B. B. Duncan, H. Nguyen, and R. Kunath, “On-Board Fiber-Optic Network Architectures for Radar and Avionics Signal Distribution,” Tech. rep., NASA/TM-2000–209776 (2000).
6. M. F. Alam, M. Atiquzzaman, B. D. Duncan, H. Nguyen, and R. Kunath, “Fibre-optic network architectures for on-board digital avionics signal distribution,” Int. J. Commun. Syst. 15(2-3), 175–190 (2002). [CrossRef]
7. F. Slaveski, J. Sluss, M. Atiquzzaman, H. Nguyen, and D. Ngo, “Trans-mission of RF signals over optical fiber for avionics applications,” in Proceedings. The 21st Digital Avionics Systems Conference, vol. 1 (IEEE, 2002), pp. 4D2.
8. N. Chand, K. McNeill, S. Kumar, and B. Eteson, “An approach to reducing SWAP and cost for avionics high-speed optical data networks,” in MILCOM 2008-2008 IEEE Military Communications Conference, (IEEE, 2008), pp. 1–7.
9. T. Antony and A. Gumaste, “Fiber-optic network architectures for on-board avionics applications investigated,” (2003).
10. S. Habiby and M. Hackert, “Motivation for WDM-based optical networks in aircraft application,” in SAE WDM LAN Task Group Meeting, (2007).
11. S. F. Habiby and M. J. Hackert, “RONIA results: WDM-based optical networks in aircraft applications,” in 2008 IEEE Avionics, Fiber-Optics and Photonics Technology Conference, (IEEE, 2008), pp. 71–72.
12. G. Proudley, N. Brownjohn, J. Baptista, A. Bacou, H. Salgado, A. Clausen, P. Deimel, M. Farries, S. Gauchy, I. Kopacek, A. Lee, B. Napier, M. Traversone, J. Vincent, and A. Zimmermann, “Developing aircraft photonic networks-an overview of the European DAPHNE project,” in 2011 IEEE Avionics, Fiber-Optics and Photonics Technology Conference, (IEEE, 2011), pp. 1–2.
13. A. Lee and S. Rayner, “Avionic architectures incorporating optical fibre technology,” in IEEE Conference Avionics Fiber-Optics and Photonics, 2006, (IEEE, 2006), pp. 10–11.
14. H. J. White, N. Brownjohn, J. Baptista, V. Foucal, H. Salgado, A. Clausen, T. Pistner, M. Farries, S. Gauchy, I. Kopacek, A. Lee, B. Napier, M. Traversone, J. Vincent, and A. Zimmermann, “Developing aircraft photonic networks for airplane systems,” in Photonic Applications for Aerospace, Commercial, and Harsh Environments IV, (International Society for Optics and Photonics, 2013), p. 87200 V.
15. A. Ragheb and H. Fathallah, “Performance analysis of next generation-PON (NG-PON) architectures,” in 8th International Conference on High-Capacity Optical Networks and Emerging Technologies,(IEEE, 2011), pp. 339–345.
16. Radio Technical Commission for Aeronautics, “Environmental conditions and Test procedures for airborne equipment”, https://do160.org/rtca-do-160g/, DO-160, RTCA DO-160.
17. Telcordia. 2006. (Kluge, R) Generic Requirements NEBS GR-63-CORE, NEBS Requirements: Physical Protection, Issue 3, March 2006, Telcordia Technologies, Inc., Piscataway, NJ. www.telcordia.com
18. K. Ennser, T. Rogowski, P. Ghelfi, F. Cugini, and P. Castoldi, “Reconfigurable add/drop multiplexer design to implement flexibility in optical networks,” in 2006 International Conference on Transparent Optical Networks, vol. 3 (IEEE, 2006), pp. 74–77.
19. T. Antony and A. Gumaste, DWDM network designs and engineering solutions (Cisco Press, 2002).
20. Y. Tian, J. Wei, M. Jiang, L. Zhang, F. Zhang, Q. Sui, and W. Sun, “Drive design and performance test of a tunable DFB laser,” in 2017 Chinese Automation Congress (CAC), (IEEE, 2017), pp. 4024–4027.
21. H. -Y. Xie, J. Li, J. -N. Gan, P. Shen, Y. -W. Huang, L. Huang, W. -R. Zhang, and W. Wang, “Complex-coupled wavelength-selectable DFB laser source for coarse WDM,” in 2008 9th International Conference on Solid-State and Integrated-Circuit Technology, (IEEE, 2008), pp. 1013–1016.
22. Y. Gu and F. Choa, “Low bias, low noise single-avalanche-stage APDs,” in The 16th Annual Meeting of the IEEE Lasers and Electro-Optics Society, 2003. LEOS 2003., vol. 2 (IEEE, 2003), pp. 1001–1002.
23. H. White, M. Watson, G. Proudley, A. Proudfoot, and N. Aldridge, “CWDM for aerospace applications-temperature testing of COTS technologies,” in 2007 IEEE Avionics, Fiber-Optics and Photonics Technology Conference, (IEEE, 2007), pp. 38–39.
