Upload
sheng-tian
View
217
Download
5
Embed Size (px)
Citation preview
Routing Space Internet Based on Dijkstra’s Algorithm
Xiang-ying Li1,2 Guo-shun Li3 Sheng-tian Zhang3
1Center for Space Science and Applied Research, Chinese Academy of Sciences Beijing 100190, China
2Graduate School of the Chinese Academy of Sciences, Beijing 100190, China 3PLA NO.63610 Koler 841001, China
Abstract—Space Based Internets will be used to move earth observation data through satellite constellations to ground based receiving stations. Routing observation traffic in this environment is constrained by individual satellite orbits, individual satellite capabilities, observation schedules, ground station schedules, and rapidly evolving natural events. Among many global routing algorithms, Dijkstra Algorithm is mostly preferred as it is faster as compared to any other algorithm and its implementation is easier. For this reason, this paper implements a routing algorithm of Dijkstra for Space Based Internets under constraints mentioned above.
I. INTRODUCTION
A Space Based Internet (SBI) is a satellite network system in which each satellite is capable of originating traffic, terminating traffic, and switching traffic traveling between other satellites and the ground. To build a SBI, each Earth observation satellite would carry a communications system with several channels (RF or optical) and a network control processor to switch traffic between the several communications channels and local payload. Fig. 1 shows a future space mission network. We expect that most switching will take place at the network layer (Internet Protocol). The advantages of a SBI in which each observation satellite takes part are several:
1) The system is scalable; as each satellite is launched it adds capacity to the SBI.
2) Standard SBI modules can be designed, constructed, and deployed to minimize individual satellite cost
3) SBI modules can be enhanced as technology advances, hence the system evolves.
4) Special SBI satellites can be designed, constructed, and deployed to provide extended capacity and capabilities.
5) Future missions are likely to be "joint" across multiple platforms and programs calling for an integrated communications capability.
6) Communications between satellites will be necessary for coordinated data collection, especially for unique events.
INTERNETEnd User
Groud Station Groud Station
RelaySatellite
Mission Satellite
ControlCenter
Fig. 1. Future Space Mission Network employing internet protocol.
As a network, a SBI is different from land-based networks in that links come up and go down based on the orbital mechanics of each satellite and the availability of line-of-sight. However, a SBI is more predictable than an ad hoc network
978-1-4244-4570-7/09/$25.00 ©2009 IEEE
because the orbits, satellite characteristics, and traffic loads are predictable. In this paper we describe the characteristics of a SBI with respect to routing, and offer initial thoughts on routing algorithms of Dijkstra suitable for a SBI.
II. STRUCTURE OF A SPACE-BASED INTERNET
There are numerous difficulties in dealing with space-based networking. Spacecraft can communicate with one another only when they are in Line of Sight, so communication links are constantly severed and re-established due to the nature of orbits. Unexpected events such as solar flares and poor weather on earth can further hamper ground-space communication. Currently, satellites do not communicate using IP, but instead use a blend of customized protocols. This communications infrastructure requires custom-built hardware and software to test and integrate the communication systems for each spacecraft. A standardized infrastructure would facilitate easier deployment, greater extensibility, and lower costs. Full support for IP-based space communication would benefit future space missions because IP is an industry standard, end-to-end communication protocol. Further, since IP is used in the Internet, it is continually being tested, researched, and developed worldwide. Therefore, NASA has developed a serious interest in applying IP to space communication networks [1] [2] [3] [4].
A SBI consists of (1) satellites equipped with a scientific instrumentation payload, a network control processor (router), and one or more communications channels, (2) ground stations to receive collected data and to manage the satellite system, and (3) optional communications only satellites to increase the capacity of the system. For the work reported here we do not distinguish between (1) and (3). Our model of a SBI satellite node is shown in Fig. 2. At the core is a packet switching system that interconnects the input/output channels and the scientific payload. The payload (generally) produces data from Earth or space observations. Once collected, the data is transferred to a ground station over the SBI. Each satellite has at least one, and perhaps several, communications channels. In our model we are not concerned with the details of these channels (e.g. radio or optical, modulation type, access mechanism, or propagation characteristics). We model communications channels as a constant bit rate link with errors and delay.
router
Communition channel
Communition channel
Communition channel
payload
Fig. 2. Model of an SBI node consists of a payload, a router, and one or more
communications channels.
We assume that the Internet Protocol (IP) is used to send, receive, and route data. Each satellite becomes a possible router in the SBI. Our task, described in the next section, is to determine the routes data should take from origin to destination.
III. SBI ROUTING
A SBI consists of a set of satellites and a set of fixed ground stations. Our task is: Given the satellite and ground station positions (current and near future) and the traffic sources and sinks, determine a set of routes from sources to sinks meeting a set of constraints. A set of routes will be in place for a reasonable length of time, perhaps 10-15 minutes or longer if the orbits allow. We will refer to the length of time a set of routes is in effect as the epoch. Satellites are in Earth orbit at altitudes at or less than geo-synchronous orbits. We assume the orbit of each satellite is well known and given the current satellite position and parameters, the satellite position in the future can be computed.
A. SBI Topology
The first step in computing a set of routes is to determine possible links between satellites and between a satellite and a ground station. Several criteria are used to determine if a link is possible:
1) Is there a clear line of sight path between the satellites or satellite and ground station?
2) Can antennas or optical telescopes be positioned to establish the path during the epoch?
3) Are the communications channels compatible (i.e. frequency, modulation, capacity, etc.)?
4) Will the path exist for a reasonable period of time, the epoch?
These criteria form an initial ‘filter’ to determine a set of possible links. The result is a graph illustrated by the sample in Fig. 3. Graph nodes are satellites or ground stations, edges are possible links, and, in this figure, satellites are position by the latitude-longitude coordinates.
Fig. 3. Illustrates the set of possible links taking into consideration
line-of-sight, compatible systems, and antenna constraints.
To determine the most useful links, we can compute the shortest path from each source to its designated sink. The more paths crossing a link, the more useful that link. In the following section we discuss the Dijkstra Algorithm which is capable for computation for the shortest path from source node to a destination one.
B. Dijkstra’s Shortest Path Algorithm in Route Computation
The problem of finding shortest paths plays a central role in the design and analysis of networks. Most routing problems can be solved as shortest path problems once an appropriate cost is assigned to each link, reflecting its available bandwidth, delay or bit error ratio, for example. There are various algorithms for finding the shortest path if the edges in a network are characterized by a single non-negative additive metric. The most popular shortest path algorithm is Dijkstra’s algorithm [5], which is used in Internet’s Open Shortest Path First (OSPF) routing procedure [6]. Dijkstra’s shortest path routing algorithm is presented below in pseudo code:
Given a network ( , )G N E= , with a positive cost ijD for all
edges ( , )i j N∈ , start node S and a set P of permanently
labeled nodes, the shortest path from start node S to every
other node j is found as follows: Initially { }P S= ,
0sD = , and j sjD d= for j N∈ ( )j S≠ .
Step 1: (Find the closest node.) Find i P∉ such
that mini jj P
D D∉
= . Set { }P P i= . If P contains all nodes
then stop; the algorithm is complete.
Step 2: (Updating of labels.) For all j P∉ , set
min[ , ]j j i ijD D D d= + , Go to Step 1.
Since each step in Dijkstra’s algorithm requires a number of
operations proportional to N , and the steps are iterated
1N + times, the worst case computation is 2( )O N [7].
Using priority queues the runtime of Dijkstra s algorithm
is 2( lg )O E N , which is an improvement over 2( )O N for
sparse networks [8]. However, the space requirement increases
and operations on priority queues are difficult to implement in
reconfigurable logic, and for these reasons priority queues
have not been dealt with in this paper.
IV. SIMULATION AND RESULTS
Once a topology has been determined, routes from sources to sinks can be determined. Most routing problems can be solved as shortest path problems once an appropriate cost is assigned to each link. Here, we take into account the link costs which reflecting its available bandwidth, delay and bit error ratio. For example, a 9-node network topology is given in Fig.4, and the costs between nodes are showed in Table 1.
0 100 200 300 400 500 600 700 800 900 1000100
200
300
400
500
600
700
800
900
1000
1
2
3
4
5
6
7
8 9
Fig. 4. Topology of network with 9 nodes
TABLE I. ILLUSTRATES COSTS BETWEEN NODES
Link Cost Link Cost Link Cost
1-3 3 2-4 7 3-9 13
1-5 9 2-7 14 4-5 17
1-6 2 2-8 6 4-8 9
1-7 8 3-4 15 4-9 11
1-9 11 3-5 6 5-9 7
2-3 11 3-7 8 6-7 14
0 100 200 300 400 500 600 700 800 900 1000100
200
300
400
500
600
700
800
900
1000
1
2
3
4
5
6
7
8 9
Fig. 5. Routing results of Dijkstra’s algorithms
Our goal is to find a shortest path with the lowest cost between node 1 and 8. After computation based on Dijkstra’s algorithm, a routing decision is made in Fig. 5 represented by red line linking nodes 1, 3, 2 and 8, the sum of cost is 20. In a centrally managed operation, topology and routes are determined at an operations center and transmitted to each satellite. The operations center can base the topology and routes on satellite position, orbital mechanics, and satellite communications capabilities.
One aspect of an SBI that differs from land-based networks is that updating the routing tables must be carefully sequenced. If a link is scheduled to go down, the source routing table must first be updated to stop transmitting on the link, and then each router along the path between source and sink must be updated, but not until all the traffic has transited the link that is going down. In an SBI this transition from one set of routes to another may take several hundred milliseconds.
V. CONCLUSION
We have characterized topology determination and routing in a Space Based Internet. Determining a topology depends on satellite location, line-of-sight, and communications channel characteristics such as bandwidth, delay and bit error ratio, which can be considered as costs in routing algorithms. Dijkstra’s algorithm has a good performance which may assist in network routing determination in space based internet.
REFERENCES
[1] Bhasin, K. and Hayden, J., “Developing Architectures and Technologies for an Evolvable NASA Space Communication Infrastructure,” 22nd AIAA ICSSC 2004.
[2] Bhasin, K. and Hayden, J., “Space Internet Architectures and Technologies for NASA Enterprises,” International Journal of Satellite Communications, Vol. 20, No. 5, 2002, pp. 311-332.
[3] Hogie, K., Criscuolo, E., and Parise, R., “Link and Routing Issues for Internet Protocols in Space,” IEEE Aerospace Conference, Vol. 2, 2001, pp. 2/963-2/976.
[4] Rash, J., Casasanta, R., and Hogie, K., “Internet Data Delivery for Future Space Missions,” Earth Science Technology Conference, NASA, 2002.
[5] E. Dijkstra: “A Note on Two Problems in Connexion with Graphs”, Numerische Mathematik, Vol. 1, 1959, pp. 269-271.
[6] J. Moy: “OSPF Version 2, RFC 2328”, May 1998. [7] D. Bertsekas, R. Gallager: Data Networks. Prentice-Hall, 1987, pp.
297-421. [8] M.A. Weiss: Data Structures and Algorithm Analysis in C. The
Benjamin/Cummings Publishing Company, Inc. 1993, pp. 281-343.