Lightweight Route Information Advertisement for LEO Mega-constellation

Introduction The continuous topology change is a main characteristic of the LEO constellation, which can be divided into the predictable topology change and the unpredictable topology change. The predictable topology change is caused by the periodic motion of the satellite, while the unpredictable topology change is caused by emergencies. With the increasing scale of satellite networks, the probability of unpredictable topology changes is also increasing. For the predictable topology change, the snapshot-based routing method divides the satellite network motion period into multiple time slices, and performs routing calculation based on time slices. Unlike the predictable topology change, the unpredictable topology change requires the corresponding protocol mechanism to capture network anomalies and flooding the route information within a certain range. However, the route information flooding without constraints would cause significant bandwidth overhead as well as additional routing convergence delay. Considering the limited on-board resources, the above method is undoubtedly inefficient. The minimum requirement to complete the route information advertisement within a certain network topology is to construct a connected graph. Therefore, the route information only needs to be advertised on the sub-topology of the physical topology. The sub-topology is constructed by removing some links from the physical topology, and has the same node reachability as the original network. The minimum spanning tree is an available sub-topology. However, the minimum spanning tree can not be used as a good sub-topology for fault-prone large-scale networks such as the LEO mega-constellation. Any failed link in these networks would decompose a tree-like structure into two disjoint sub-topologies, and prevent information advertisement. In addition, with the increasing scale of the network, the time complexity of the algorithm to construct the sub-topology is furhter increased. Aiming at the above issues, this document provides a lightweight route information advertisement mechanism in the LEO mega-constellation. On the one hand, the mechanism selects the route information advertisement link by the way of route-associated judgment to reduce the overhead of route information advertisement. On the other hand, the mechanism provides a method for dealing with link fault during the route information advertisement process, to ensure the reliability of routing information advertisement.

Terminology

VLEO: Very Low Earth Orbit with the altitude below 450 km.
LEO: Low Earth Orbit with the altitude between 180 km and 2000 km.
MEO: Medium Earth Orbit with the altitude between 2000 km and 35786 km.
GEO: Geosynchronous Orbit with the altitude 35786 km.
Intra-satellite links: Links between adjacent satellites in the same orbit.
Intra-satellite links: Links between adjacent satellites in the different orbits.
SGP4: Simplified Perturbations Models.
BGP: Border Gateway Protocol [RFC4271].
IGP: Interior Gateway Protocol, examples of IGPs inculde Open Shortest Path First (OSPF [RFC2328]), route information Protocol (RIP [RFC2453]), Intermediate System to Intermediate System (IS-IS [RFC7142]) and Enhanced Interior Gateway Routing Protocol (EIGRP [RFC7868]).

Basic Network Structure In the LEO mega-constellation, each satellite moves along the orbit, which can be divided into circular orbit satellites and elliptical orbit satellites. Different orbits can be described by Keplerian parameters. At present, the mainstream of satellite networks basically adopt circular orbit. When links between satellites are established for end-to-end communication, each satellite usually has a fixed number of links which communicate with neighboring nodes, and considering the cost of satellite links and power restrictions of satellites, satellite links are generally limited to direct connections between adjacent nodes. In a single-layer satellite constellation, each satellite has four types of contiguous neighbour satellites and each type refers to a direction. The number of neighbor satellites distributed in one direction is determined by the number of antennas deployed on the satellite for communication. As shown in Figure 1, if the satellite contains a single antenna in each direction, the satellite Node E has two links in the same orbit and two links in different adjacent orbits with other satellites. In a multi-tier satellite constellation, each satellite may have two additional types of adjacent satellites, upper level satellites and lower level satellites in different tiers.

Node E and its adjacent satellites The link delay is the main factor that affects the advertisement efficiency. The longer the inter-satellite distance is, the longer the advertisement propagation delay is. The propagation delay is related to the accumulated delay of upstream links and the link delay in the basic network structure. Figure 1 shows the basic network topology which is managed by Node E, including 8 surrounding nodes of Node E and links connecting these nodes. Since the satellite moves regularly along the orbit, each node can maintain the propagation delay of edges in the basic structure based on its own position and orbit parameters. The propagation delay of the edge can be marked as W, e.g. W_AB. Based on the accumulated delay of upstream links and the link delay in the basic network structure, each node can realize the route-associated judgment of information advertisement. It should be noted that the basic structure in the real network is not presented as a simple rectangle or square for the variation distances between nodes.

Route-associated Judgment As shown in Figure 1, Node E has four links connected to the neighbor nodes. If the routing information is announced from one of these four links, Node E judges whether to advertise the route information to other neighbors based on the accumulated link delay and the link delay in the basic network structure. The route information advertisement is classified into the horizontal direction advertisement and the vertical direction advertisement. According to the receiving direction of the route information, the information advertisement is divided into the fault related advertisements and the non-fault related advertisements. The fault occurs in different network locations can be divided into the horizontal direction network fault and the vertical direction network fault.

Horizontal direction network fault

Vertical direction network fault | | +---+ +---+ | +---+ +---+ --| |----| A |-|--| C |----| |-- +---+ +---+ | +---+ +---+ | X | | | | | | | | +---+ +---+ | +---+ +---+ --| |----| B |-|--| D |----| |-- +---+ +---+ | +---+ +---+ | | --+--> | | | | | | | +---+ +---+ | +---+ +---+ --| |----| |-|--| |----| |-- +---+ +---+ | +---+ +---+ | | | | | Left Part | Right Part ]]> According to different fault types, the route information advertisement manner at the source node should be classified and discussed. For a network fault in the horizontal direction, the whole network topology is divided into a left part and a right part by the vertical plane of the failed link as shown in Figure 2. The source node advertises information from links except the failed link, and all route information is not advertised across the left part and the right part. For a network fault in the vertical direction, the network topology is divided into the left part and the right part by two adjacent orbit planes where source nodes and corresponding neighbor nodes are located. The source node advertises information from links except the failed link. It should be noted that except for the first hop advertisement in the horizontal direction, the route information is not advertised across the left or right part. As shown in Figure 3, Node C and Node D are the right neighbors of the source nodes A and B of the failed link. And the network topology is divide into a left part and a right part by the orbit planes of Node C, D, A, B. 1) Route Information From the Horizontal Direction

Route information from the horizontal direction When the route information is advertised from the horizontal direction, the upstream node should inform the advertised node of the accumulated link delays of itself and its adjacent nodes at the same orbit. After receiving the route information and the accumulated link delays, the advertised node judges whether to advertise the route information to the downstream node according to the network information maintained by itself. As shown in Figure 4, the advertised node informs the downstream node in the horizontal direction by default. The basic process is as follows: Step 1: The route information and the accumulated link delays are advertised to Node E by Node F. The accumulated link delays refer to the delays when the routing information reaches Node C, F, and I, which are marked as Sum_C, Sum_F, and Sum_I. Step 2: After receiving the route information, Node E needs to determine whether to advertis the route information to the downstream nodes, including Node B, D, and H. For Node B: Node E evaluates the accumulated link delay at Node B from the perspectives of Node F and Node C respectively, denoted as Sum_FB=Sum_F+W_EF+W_BE and Sum_CB=Sum_C+W_BC. If Sum_FB is less than Sum_CB, Node E advertises the route information to Node B, otherwise Node E does not advertise. For Node D: Node E advertises the route information to Node D by default, and simultaneously informs the accumulated link delay when the route information reaches Node B, E, and H, which are marked as Sum_B, Sum_E, and Sum_H. 2) Route Information From the Vertical Direction (1) Fault Related Link Scenario When the route information is advertised from the vertical direction link which is a fault related link, the upstream node advertises the route information from the vertical direction by default, and informs the advertised node the accumulated link delay of itself. After receiving the route information and the accumulated link delay, the advertised node continues to advertise to the downstream node in the horizontal direction by default. As shown in Figure 5, Node K is a node at the fault related link. The basic process is as follows:

Fault related link scenario >>>> | +---+ +---+ +---+ +---+ --| E |----| F |----| G |----| H |-- +---+ +---+ +---+ +---+ | | ^ | | | | ^ | | +---+ +---+ +---+ +---+ --| I |----| J |----| K |----| L |-- +---+ +---+ +---+ +---+ | | ^ | | ^ Routing ^ Info ]]> Step 1: The route information and the accumulated link delay are advertised to Node G by Node K. The accumulated link delay refers to the delay when the route information reaches Node K, which is marked as SUM_K. Step 2: According to Sum_K and the link delay maintained by Node G, Node G informs Node F of the accumulated link delay estimations when the route information reaches Node C, G, and K, which are recorded as Sum_C, Sum_G, and Sum_K. Sum_C and Sum_G could be denoted as Sum_C=Sum_K+W_CG+W_GK and Sum_G=Sum_K+W_GK respectively. Step 3: After the route information reaches Node F, the route information advertisement method of Node F regresses to the process described in 1). As mentioned earlier, the route information is not advertised across the left part and right part. When the route information is advertised from the vertical direction link which is a fault related link, the route information would only be advertised to the neighbor nodes at one side in the horizontal direction. There is no need to limit the notified side, and the notified side may be unified in advance when performing the associated judgement of the route information advertisement. As shown in Figure 6, if Node G advertises routing information to Node F, it would not advertise to Node H. Except for special cases, when a link fault occurs in the vertical direction link, the source node will advertise to the neighbor nodes on both sides in the horizontal direction. (2) Non-fault Related Link Scenario

Non-fault related link scenario When the route information is announced from the vertical direction link which is a non-fault related link, the upstream node announces the route information in the vertical direction by default, and does not inform the advertised node of the accumulated link delay of itself. After receiving the route information, the advertised node continues to inform the downstream node along the vertical direction by default. The downstream node continues to advertise the route information in the vertical direction until the arriving node has received the same route information previously. If a node has received a route information, the subsequent same route information will not be announced to the downstream node. Meanwhile, the route information announced by the non-fault related link along the vertical direction will not be announced along the horizontal direction. As shown in Figure 6, Node D, H, and L advertise route information as upstream nodes. The basic process is as follows: Step 1: After receiving the route information from Node D, Node C judges based on the process described in 1) and advertises the route information to Node G and Node B. Meanwhile, Node H advertises to Node G by default. Step 2: For Node G: After receiving the route information from Node C, since Node G is in the non-fault related link and Node C advertises to Node G in the vertical direction, Node G continues to advertise the route information to Node K along the vertical direction where the link between Node C and Node G is located by default. At the same time, relevant messages are no longer announced to Node F. The subsequently received route information advertised by Node H will be discarded by Node G. Assuming that Node K has received the information at Node L in advance, Node K decides to announce the route information to Node J based on the process described in 1). Meanwhile, the subsequent route information from Node G will be blocked at Node K. Step 3: For Node B: The route information and the accumulated link delay are announced to Node B by Node C. Since the horizontal links where Node B and Node C are located have a lower delay than the horizontal links where Node F and Node G are located, the subsequent nodes of the horizontal link where Node B and Node C are located will advertise the route information downward by default. After receiving the advertisement of Node C, Node B decides to advertise the route information to Node A and Node F according to the process described in 1). Step 4: For Node F: After receiving the route information, Node F continues to inform Node J. If the information transmitted by Node F arrives earlier than the information at Node K, Node J continues to announce along the vertical direction link, and at the same time, the subsequently received message sent by Node K to Node J is no longer announced to Node I.

Error Handling The above method can ensure the synchronization of the route information among the nodes in the certain network range. However, the single point fault would cause the synchronization error of route information during the advertisement process. There are two types of single point fault, including the horizontal fault and the vertical fault. In this section, the fault handling method is discussed. It should be noted that if a link affected by a single point fault is not the link selected by the route-associated judgment, the route information advertisement process is not affected by the single point fault, as shown in Figure 7.

Horizontal direction fault without impact 1) Handling Method For the Horizontal Direction Fault When the vertical route information arrives later than the horizontal route information or there is no vertical route information advertisement, the single point fault would affect the route information advertisement process. In this case, the detour link with the smaller total delay is selected to advertise the route information. In order to ensure the consistency of the accumulated link delay, the detour link delay is not added to the accumulated link delay, and the accumulated link delay in the normal state is still used as the basis for determining the route information advertisement of the downstream node.

Horizontal direction fault with impact As shown in Figure 8, the link between Node D and Node E is interrupted. When the route information arrives at Node E, the route information would pass through Node B and Node A in turn which has lower path delay compared with the other detour paths. At the same time, Node E would inform Node D the accumulated link delays at Node B, E, H. 2) Handling Method For the Vertical Direction Fault The vertical direction fault during the advertisment process could be divided into two scenarios, including the non-fault related link and the fault-related link. The fault handling methods for these cases are described below. (1) Fault Related Link Scenario In the fault related link scenario, the route information advertised from the upstream node would bypass to one side at the fault point and reach the adjacent orbit. The advertisement process would be restarted at the adjacent orbit, as shown in Figure 9. The basic process is as follows:

Fault related link scenario > +---+ ---| A |-----| B |-----| C |--- +---+ +---+ +---+ | ^| | | ^| | +---+ << +---+ >> +---+ ---| D |-----| E |-----| F |--- +---+ +---+ +---+ | ^| | | ^| X +---+ << +---+ << +---+ ---| G |-----| H |-----| I |--- +---+ +---+ +---+ | | ^| ^ Routing Info ]]> Step 1: For the link fault between Node I and Node F, Node I informs the route information and the acculumated link delay Sum_I from the neighbor orbit. Step 2: After Node H receives the route information and Sum_I, it continues to inform Node G by default. At the same time, Node H advertises the route information and Sum_H to Node E in the vertical direction. Sum_H could be represented by following Formula. Sum_H=Sum_I+W_HI Step 3: After Node E receives the route information and Sum_H, it continues to inform Node D by default. At the same time, Node E advertises the route information and Sum_E to Node B in the vertical direction. Sum_E could be represented by following Formula. Besides, Node E advertises the route information to Node F by default. Sum_E=Sum_H+W_HE Step 4: Node B and subsequent downstream nodes repeat the above process until a downstream node in the vertical direction which has received the same route information. (2) Non-fault Related Link Scenario In the non-fault related link scenario, the route information from the upstream node would detour to the unadvertised downstream node side at the fault point, as shown in Figure 10.

Non-fault related link scenario > +---+ +---+ << ---| D |-----| E |-----| F |--- +---+ +---+ +---+ v| v| v| v v v ]]> Step 1: For the link fault between Node B and Node E, Node B announces the route information from the neighbor orbit. Step 2: Node A continues to inform Node D after receiving the route information. Step 3: After receiving the route information, Node D informs Node E, so as to complete the detour process of the route information advertisement.

Route Information Advertisement As shown in Figure 11, a fault occurs in the link between A1 and A2 at a certain time. The network topology is divided into the left part and the right part by the vertical plane of the fault link. For the left part, starting from the link fault point A2, the route information is advertised to the neighbor nodes in the three direction of up(B2), down(E2), and left(A3). The specific process is as follows:

Route information advertisement |<----->| Left Part Right Part ]]> 1) Advertisement Process From A2 to A3 The advertisement process from A2 to A3 is as follows. Step 1: A2 informs A3 of the route information and the accumulated link delays which include Sum_B2, Sum_E2, and Sum_A2. In combination with its own position and orbit parameters, A2 regularly maintains the network information in the basic structure. The accumulated link delays could be represented by following formulas. Sum_B2=W_A2B2 Sum_E2=W_A2E2 Sum_A2=0 Step 2: After receiving the route information, A3 needs to determine whether to announce the route information to B3, A4, and E3. If Sum_A2+W_A2A3+W_A3B3 is less than Sum_B2+W_B2B3, A3 advertises the route information to B3, otherwise, no advertisement is required. If Sum_A2+W_A2A3+W_A3E3 is less than Sum_E2+W_E2E3, A3 advertises the route information to E3, otherwise, no advertisement is required. By default, A3 informs A4 of the route information and the accumulated link delays which include Sum_B3, Sum_E3, and Sum_A3. In combination with its own position and orbit parameters, A3 regularly maintains the network information in the basic structure. The accumulated link delays could be represented by following formulas. Sum_B3=Sum_B2+W_B2B3 Sum_E3=Sum_E2+W_E2E3 Sum_A3=W_A2A3 The process of above route information advertisement from A3 to A4 is repeated until the edge node of the left part is reached or the route information on the vertical direction is received. 2) Advertisement Process From A2 to B2 Step 1: A2 informs B2 of the route information and the accumulated link delay corresponding to the arrival of the route information at A2 which is marked as Sum_A2=0. Step2: After receiving the route information, B2 continues to advertise the route information as well as the accumulated link delays to C2 and B3. For Node C2: The advertisement process from B2 to C2 is consistent with the scenario summarized in the previous section. B2 informs C2 of the accumulated link delays. The route information advertisement process from B2 to C2 continues to repeat and reach the D2 in the near-polar region of the western hemisphere. The route information passes through the polar region until it reaches the node at the semi-perimeter position of the same orbital plane. For Node B3: The advertisement process from B2 to B3 is consistent with the scenario summarized in the previous section. B2 needs to inform B3 of the accumulated link delays which include Sum_A2, Sum_B2 and Sum_C2. In combination with its own position and orbit parameters, B2 regularly maintains the network information in the basic structure. The accumulated link delays could be represented by following formulas. Sum_A2=0 Sum_B2=W_A2B2 Sum_C2=Sum_B2+W_B2C2 Step 3: The route information announced by C3 reaches C4 after multiple hops, and C4 makes a judgment according to the horizontal link announcement process summarized in the previous section, and announces the route information to B4. Step 4: After the B4 receives the vertical route information, on the one hand, the subsequent horizontal advertisement received by the B4 will be stopped, and the information will not be advertised to the next-hop horizontal node B5, on the other hand, the B4 advertises the information to the A4 along the vertical direction, and if the A4 has received the route information, the advertisement will be stopped, otherwise, the advertisement will continue. 3) Advertisement Process From A2 to E2 The advertisement process from A2 to E2 is the same as the advertisement process from A2 to B2, and the route information advertisement process on the right side of the vertical line is the same as the route information advertisement process on the left side of the vertical line, which is not described here.

Future Works In this document, the method of route information advertisement in the LEO mega-constellation is discussed. It should be noted that the satellite network is one of the application scenarios. The LEO mega-constellation network constructs a mesh topology. For other scenarios which have the same network topology property, this method could also be applied. In the future work, the extension of the current routing protocol to support the method of route information advertisement described in this document would be taken in mind.

Security Considerations TBA.

Acknowledgements TBA.

IANA Considerations This document has no IANA action requested.