Communications, navigation and surveillance services in the National Airspace System (NAS) of the future will reflect significant advances in technology and capability, and in the distribution of responsibilities for air traffic management. Digital communications will be the cornerstone of next generation systems aimed at increasing transportation access and mobility, alleviating current airspace-capacity saturation. Such a network will likely be based on the Internet Protocol and will have to support the inherently mobile nature of aircraft. This paper presents the concept of self-configuring ad-hoc wireless networks that, when combined with existing wired network technologies, can provide self-configuring, self-healing, fault-tolerant communications for mobile nodes that must also communicate with fixed, wired network infrastructure. This hybrid network architecture enables mobile nodes to collect and process data both independently and cooperatively as opposed to the centralized HOST data processing approach currently used for air-traffic coordination.

Challenge: As the U.S. transportation system nears its capacity limits [I], alternative operational models and intelligent transportation systems will be employed to satisfy the national public demand for increased access and mobility and to increase the current transportation system capacity.

In an examination of airspace system capacity, Dr. George Donohue [2] and others suggest that reduced aircraft separation and increased capacity at major hubs will not produce enough additional capacity for projected traffic growth. His projections of "hub and spoke" saturation imply that increased utilization of smaller airports will be essential to maximize air transportation system capacity.

The FAA is addressing the capacity issue with, among other initiatives, Free Flight; a concept that will utilize data link in an integrated network of air, ground, and airborne communications systems [3]. NASA's Small Aircraft Transportation System (SATS) goes even further. This NASA/FAA/industry research partnership will demonstrate a distributed, point to point transportation system serving all of the nation's 18,000 underutilized landing facilities in near all weather conditions. The new access to rural communities and direct links for trips up to 1500 miles this program proposes could significantly increase capacity and reduce dependence on the fewer than 600 "hub and spoke" airports through which 96% of today's air travelers pass. Emerging communications, navigation and surveillance (CNS) technologies and what NASA refers to as the "airborne Internet" will change the public transportation paradigm and create new challenges in air traffic management (ATM), (http://sats.nasa.gov/)

Effectively, the current method of air-traffic control is at its capacity limits, and while procedural changes such as separation reductions may provide short term interim solutions, programs like the ones above show technology is being sought to provide a longer-term solution. The heart of any technology solution will be digital communications links, i.e., the integrated network[3], between aircraft and transportation facilities.

We assume an Internet Protocol (IP) network will be employed to leverage the vast amount of existing commercial-off-the-shelf (COTS) hardware and software that supports IP. Over the past 30 years, the protocol has proven itself capable of adapting to a myriad of environments from low-bandwidth wireless links, emerging gigabit networks, and high-latency satellite environments.

The problem, however, is developing an IP-based architecture that meets the requirements of the air-traffic environment. The environment would likely employ conventional wired, local-area network (LAN) and wide-area network (WAN) components for facility interconnection, but it is aircraft mobility that presents the challenge. Effectively, today's air-traffic control represents a hybrid wireless/wired network environment with high wireless-node mobility.

What needs to be done. A hybrid wired/wireless network should be considered for NAS. Traditional wire-line networks are built on the assumption that nodes remain relatively static over time. That is, once an IP address is assigned to a node, that address is relative to the local sub-network in which it is installed, and a default router on the same sub-network provides the connection to the all other sub-networks.

Self-organizing ad-hoc networks attempt to remove the static-hierarchy constraints of traditional wire-line networks by implementing mechanisms whereby a node's physical and logical placement in a network doesn't matter from the standpoint of providing connectivity to other nodes.

Since aircraft are inherently mobile, an IP-network architecture that considers the mobility of a large percentage of network nodes needs to be developed. Self-organizing ad-hoc networks can have an important role in the resultant architecture for the following reasons:

• support for mobile nodes and automatic reconfiguration

• significant reduction in installation and maintenance costs

• built-in fault-tolerance and self-healing mechanisms

In a self-organizing ad-hoc network, a node would be free to move anywhere throughout the environment without any user intervention to obtain a new address, and there is no longer a concept of sub-networks with related IP addresses and network masks. In fact, the determining factor of when to find a new route is often driven the transmission range of the wireless network interface as opposed to a pre-determined hierarchy.

Obviously, an aircraft traveling across the country could make use of self-organizing ad-hoc network technology to maintain IP connectivity with other aircraft as well as ground-based air-traffic control computers.

What has been done/what can be leveraged. Routing protocols used in conventional wire-line environments fall into two categories: those based on link-state protocols and those based on distance-vector protocols. Examples of link-state routing protocols include the replacement for the original ARPANET routing protocol [4], IS-IS (adopted by the ISO) [5], and OSPF [6]. Examples of distance-vector routing protocols include RIP [7], IPX [8], and Xerox's XNS [9]. For various reasons wire-line-oriented protocols are not suitable for wireless LANs.

The Mobile and Ad Hoc Networking (MANET) working group of the Internet Engineering Task Force is currently investigating a number of routing algorithms for ad-hoc wireless networks. The candidate protocols include the Ad Hoc On-Demand Distance Vector (AODV) protocol [10], the Temporally Ordered Routing Algorithm (TORA) [11], the Zone Routing Protocol (ZRP) [12], and the Dynamic Source Routing (DSR) protocol [13]. AODV, TORA, and ZRP are all designed for networks with bi-directional links; they may produce invalid routes in networks with unidirectional links.

In contrast to other MANET protocols, DSR is designed to operate correctly in networks with unidirectional links. The drawback of DSR is its residual overhead associated with routing data packets. As the length of the route increases, the size of the data packet also becomes larger. When compared to hop-by-hop routing, the source-routing approach used by DSR incurs larger steady-state routing overhead in terms of the CPU cycles used and the power consumed by a node to forward a packet.

At Architecture Technology we have implemented a nomadic-routing protocol known as the Source-Initiated Adaptive Routing Algorithm (SARA) that, among other features:

• works in bi-directional and unidirectional environments

• can automatically detect the emergence of unidirectional links in a previously bi-directional environment and automatically change operation to accommodate the unidirectional links

• attempts to minimize residual overhead associated with routing packets

SARA is currently implemented using Lucent 802.11 WaveLAN wireless Ethernet interfaces on both the Windows NT and Linux platforms; a subset of the protocol has also been adapted to an embedded robotic environment that uses low-bandwidth RF links.

The cluster gateway contains both wireless- and wired-network interfaces. The multiple network interfaces coupled with mechanisms based on the Mobile IP specification [14] is what allows the wireless nodes to be located by all other nodes in the network regardless of where a wireless is currently located or where it moves to in the future. The cluster gateway is currently being implemented to support Lucent 802.11 WaveLAN cards and Ethernet networks though other wireless and wired networks could be supported.

Of course, there is no free lunch. The very mechanisms that enable seamless connectivity in a self-organizing ad-hoc network also consume available bandwidth. Furthermore, security is also a concern as every node is potentially a router, and rogue nodes can easily pretend to be routing packets that they are in fact consuming or sending to erroneous (a situation referred to as "route hijacking"). However, the benefits of node mobility generally outweigh bandwidth concerns, and there are several initiatives underway, from Architecture Technology and others, that address the issue of security in wireless environments.

Benefits to the nation. The implication of doing nothing to improve current national airspace (NAS) capacity is drastic, suggesting that the status quo of surface gridlock and air "hublock" will constrain the nation's economic and social growth. From the perspective of implementing a ubiquitous IP infrastructure to aircraft and air-traffic facilities, the impact is a bit more subtle: technological gains that can be made by distributing the computational processing loads to new intelligent aircraft and pushing preliminary decision making to the fringes of the network (i.e., the aircraft) will go unrealized. Not using technology to alleviate the congestion problem will likely lead to interim solutions that only provide a brief respite from current traffic saturation.

The flexibility of a hybrid network that integrates wireless-node mobility with established and emerging wired-LAN and -WAN technology is ideally suited to next-generation air-traffic management environments. Aircraft in a particular geographic area could communicate directly with one another, with nearby surface air traffic management, flight or commercial information facilities, or with any other aircraft or ground-based node within the entire NAS infrastructure. Furthermore, this system could be incrementally implemented at a significantly lower cost than, say, a satellite based system, while providing significantly greater bandwidth and scalability.

This represents but a first step toward a National Transportation system which makes a shift in travel and product distribution speeds comparable to that achieved early in the last century by the transition from horse to rail, or rail to automobile. The economic and social implications of such a change in speed are enormous. Studies, such as one commissioned by the State of Michigan, suggest that the economic value of converting a single visual airport to accommodate instrument operations results in about $10 Million in new revenue for local communities. The communications scheme proposed in this paper might be an essential element in making near all weather access to such airports possible, enabling economic development and wide realization of such benefits.

References:

1. Mineata, N. "Avoiding Aviation Gridlock: A Consensus for Change," National Civil Aviation Review Commission, 1997. (http://www.awgnews.com/faa/faa.htm)

2. Donahue, G.L. "A Simplified Air Transportation System Capacity Model," Journal of ATC, pp. 8-14, April-June 1999.

3. FAA Web site page, "Federal Aviation Administration Free Flight Introduction," http://www.faa.gov/freeflight/ff_ov.htm

4. J.M. McQuiIlan and D.C. Walden, "The ARPA Network Design Decisions," Computer Networks, Aug. 1977.

5. International Standards Organization, "Intermediate System to Intermediate System Intra-Domain Routing Exchange Protocol for Use in Conjunction with the Protocol for Providing the Connectionless-Mode Network Service (ISO 8473)," ISO DP 10589, Feb. 1990.

6. J. Moy, "OSPF version 2," Internet Request for Comments RFC 1247, Jul. 1991

7. C. Hedric, "Routing Information Protocol," Internet Request for Comments RFC 1058,Jun. 1988.

8. P. Turner, "NetWare Communications Processes," NetWare Application Notes, Novell Research, Sept. 1990.

9. Xerox Corporation, "Internet Transport Protocols," Xerox System Integration Standard 028112, Dec. 1981.

10. C. Perkins "Ad Hoc On Demand Distance Vector (AODV) Routing," Internet Draft (work in progress), Nov. 1997.

11. V. Park, and S. Corson, "Temporally-Ordered Routing Algorithm (TORA) Version 1 Functional Specification," Internet Draft (work in progress), Nov. 1997.

12. Z. Haas, and M. Pearlman, "The Zone Routing Protocol for Ad Hoc Networks," Internet Draft (work in progress), Nov. 1997.

13. J. Broch, D. Johnson, D. Maltz, "The Dynamic Source Routing Protocol for Mobile Ad Hoc Networks," Internet Draft (work in progress), Mar. 1998.

14. IETF RFC