MegaReduce

MegaReduce is a project investigating an important research problem facing the upcoming satellite Internet: from a network perspective, how many satellites exactly do we need to construct a survivable and performant LSN?

This repo contains the source codes of MegaReduce and examples to demonstrate its usage in optimizing the incremental deployment and long-term maintenance of future satellite Internet.

Working principle

This project can achieve an automatic reduction of constellation size and avoid NP-hard problems.

Given the initial configuration of a constellation, the constellation size is optimized after being checked for survivability and performance requirements

Code Introduction

The working environment needs to use python 3 version, you just needpython3 MR_demo.py to run this code.

Change your requirement and set your input in configuration.json

Each file's function

• Topo.py and Constellation_model.py describe how communication take place in satellite networks and generate the data structure of the graph.
• utils.py provide Coordinate transformation, graph algorithm and methods for finding disjoint paths.
• gw_amazon.json and gw_starlink.json describe the location of ground stations of each satellite network.

how to use demo?

Modify configuration.json

1. Specify the initial constellation configuration assign init_con,each item in list means [Shell number, altitude, orbital inclination, number of orbits, number of satellites per orbit] of a constellation respectively.
2. Figure out the file path of ground stations
   By default,we provide the potential ground stations' locations in first half of 2023.
3. Specify delay constraints, which means The hop count of the optional path should be less than the value multiplied by the shortest hop count.
4. Select the geographical location of the communication point. Make them evenly distributed across the globe.
5. Specify the Survivability requirement survivability,which means the disjoint path between any two communication pairs should not be less than this value.
6. In order not to take too long, we set the number of iterations to no more than ten. if (shrink_count + expand_count) > iterations and ret as condition for the end of the program.

The iteration result is shown in con_log.log

Modify the shrink/expand algorithm in MR_demo.py

1. Design expansion and shrinking strategies for constellations in function con_expand and con_shrink. We use the method described in paper by default.You change the step-size,even this method to fit your requirement.

2. If you want to simulate the random damage or concentrated damage, uncommenting the function stochastic_down orsun_storm_emulation in Constellation_model.py.