JPH0456442A - Method for routing flooding packet to node - Google Patents

Abstract

PURPOSE: To make information management easier by eliminating unnecessary recirculated packets by making rebroadcasting only when the multi-address communication or a flooding packet received at a node meet a specific criterion. CONSTITUTION: In the grid of packet nodes 10-30, each packet is indicated by the circle around 'X'. At each node, a terminal node controller having independent processing and storing abilities is provided. In addition, links 32-80 are provided between selected nodes. The number in the circle at the end of each link indicates a link sequence. Each local node only rebroadcasts a flooding packet to a remote node having priority and a lower link sequence. Each node exchanges information with another through a packet at the beginning of communication for informing another node of the link sequence number. This constitution can make information management easier, because unnecessary recirculated packets are eliminated.

Description

BACKGROUND OF THE INVENTION This invention relates to packet communications, and more particularly to methods for sending large numbers of packets in complex packet-switched communications networks.

One of the applications for packet-switched communication networks, particularly broadcast networks, is the remote polling of a large number of stations or nodes at any location. A particular application is remote reading of power meters in metropolitan areas, where each meter is separately addressable. Such a network may include multiple nodes. Until now, there have been no networks of such size and complexity.

Unlike central office switching systems, such as telephone systems, packet switching systems do not rely on central control. Therefore, successful communication of packets between nodes is a function of contention. Success is inversely proportional to the amount of traffic on the channel. Contention makes traditional packet switching techniques particularly difficult to use in broadcasting messages to all nodes in large networks.

The conventional packet routing protocol used to broadcast messages to all nodes in a network is the "flooding sink" algorithm. Under the flooding sink algorithm, packets are broadcast by the source to all nodes in all directions within the source's reception area, and then each of the receiving nodes and rebroadcast the packet to all nodes in all directions. The protocol can be likened to ripples in a pond caused by the splashing of a single stone, which also causes further ripples from objects on the surface of the pond. This process repeats until all nodes have received the second copy of the packet and all copies of the packet cease to circulate.

One technique used to suppress packet circulation is to prevent a node from rebroadcasting a packet if it has been previously received. Thus, once every node has received two copies of the same packet, the packet will stop circulating.

Flooding protocols work well in sparsely populated networks. Recirculation of the same packets in such systems does not overload the network. However, if there are more than a few hundred nodes in the network, contention problems between recirculating packets could cause communication overload and effective breakdown of the packet network.

Two solutions other than the current invention have been proposed to solve the overload problem in large networks.

The first proposed solution is to selectively but randomly address small groups of nodes in the receiving area. Although this reduces the number of packets in circulation, as explained below, it also works in the opposite direction of this invention and allows packets to be synchronously distributed throughout the network without addressing the fundamental problem. report. A second proposed solution is to have the source node poll each node and address each node individually.

However, such an approach incurs more overhead in managing the information than it manages. In a large network,
Management information may overwhelm message information.

The trade-off between omnidirectional flooding, with its chaotic information propagation, and polling, with its bureaucratic management overhead, is that each of them results in an eventual overload of the network, but in large network designs This presents a dilemma that remains unresolved to date.

The hope is to provide a protocol with low administrative overhead that eliminates unnecessary recirculating packets and does not create a splitting management problem, while preparing for the timely arrival of every packet of a message. .

SUMMARY OF THE INVENTION In accordance with the present invention, a technique is provided in which broadcast or flooding packets received at a node are rebroadcast only if certain criteria are met. Packets received over a communication link where sequence information is maintained at the node on a link basis for each received packet and are assigned a sufficiently high priority, i.e. a sufficiently low sequence number. The criterion is based on a packet routing protocol where the packets are rebroadcast only to other nodes. Each node has sufficient information to maintain information about each other nodes within its communication range and to assign link sequence numbers to linked nodes upon the first exchange of link information packets. Each node works backwards from the information sent to it to determine which other nodes should receive the rebroadcast packet. In a packet flooding situation, a node rebroadcasts flooded packets only to those other nodes that have low sequence numbers or for which the rebroadcast node is related by the low sequence numbers of other nodes. Flood packets leave the source node to other nodes,
The protocol is further modified by indicating that rebroadcasts may only be rebroadcast to nodes located in a far semicircular region, i.e. bisected by an axis between the source node and intermediate rebroadcast nodes. Good too.

The invention will be better understood with reference to the following detailed description and drawings.

DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS Referring now to FIG. 1, packet node 10.12
．． 14.16.18.20.22.24.26.28,
and 30 grids are illustrated.

Each of the nodes is represented by a cross or a circle around an X. At each node in the packet system, a terminal node controller (not shown) is provided with independent processing and memory capabilities.

Between the selected nodes, link 32.34.36.38
．． 40,42,44,46.48.50゜52.54,
56, 58. 60. 62. 64. 66, 68. 70
．． 72. 74. 76, 78. and 80. Links are represented by solid or dashed lines, the meaning of which will be explained in detail. Each link is also illustrated with a directional arrowhead indicating the preferred direction of propagation, as explained in Dinghai. At the end of each link there is a number in a circle. The enclosed numbers represent the link sequence for the associated nodes and links. A link sequence is a number assigned by a node to each link that each node is within its communication range.

In accordance with the present invention, each local node has a priority higher than a preselected threshold, i.e., a local node with a priority assigned by a remote node to a local node with a lower link sequence (e.g., 1.2, or 3). , rebroadcast the flooding packets only to those remote nodes. Each of the nodes exchanges information via packets at the beginning of communication between the nodes in which the nodes inform each other of the link sequence numbers for the links established by it. For example, each node in FIG. 1 has links with other nodes having the following link sequence. The asterisk indicates that under global flooding, the flooded packet is rebroadcast across the link between two tabulated nodes of the tabulated link (if the remote node is not the source node) In the table below, the local link sequence is the priority assigned by the local node to the link leading to the remote node, and the remote link sequence is the priority assigned by the remote node to the link at the local node. It is gender.

(Left below) Table I Link sequence assignments may be assigned in the order in which links are established at a node. The first link to be established is assigned the lowest link sequence. For example: Alternative assignments may be made without departing from the scope of this invention, such as link sequence numbers based on experience in the quality of the communication links.

The flooding protocol may be modified to limit directionality. For example, the flooding protocol may limit rebroadcasts in a direction away from the source node, more precisely in a direction in a semicircle facing the source link. Additionally, the flooding protocol may limit flooding in the compass direction with respect to the primary source. Thus, in FIG. 1, the direction of propagation of the flooding packet rAJ is from node 10 generally north (top of the page), i.e., south, as indicated by the arrow at one end of each link. There are no elements.

FIG. 1 illustrates the propagation of a single flooding packet rAJ, where the number of links is indicated by the suffix. Packet AO indicates the first link, packet A1 indicates packet A
packet A2 is the first rebroadcast of packet AO
This is the second rebroadcast of. Each of the packets is local (
node to node) are the same except for addressing.

Standard acknowledgments used in packet protocols may be used to confirm receipt of the packet.

Although many links are established between nodes, only a selected number of links are used to rebroadcast flooded packets. Unused links under flooding of packets are illustrated as dashed lines in FIG. 1 in the protocol, where the highest allowed link sequence is 3 at both the local and remote nodes.

The redundancy of such an arrangement will be obvious. Nevertheless, the protocol excludes recirculating packets.

The link illustrated in FIG. 1 is a simple one-hop link. More complex links may also be defined through intermediate nodes. Complex links may be assigned link sequences in the same manner as simple links.

Directivity of flooded packets is established by restricting rebroadcasts to nodes far from the source node. Depending on the design of the network, directivity may be established by knowledge of the location of the source node. This is most easily done by embedding the coordinate location of the source node within the address of the originating source node. This allows the local node to calculate compass direction based on its own location and its knowledge of the location of remote nodes with which it has communication links. The imaginary axis passing through the starting source node (10 in FIG. 1) and the first intermediate node 12 may serve as a baseline.

FIG. 2 is a flowchart illustrating the operation of sequence number acquisition according to the present invention. This operation occurs within the terminal node controller at each node. The first step is to receive the packet (step A). The packet is then tested to determine whether the packet is from a node with which it has not previously communicated (step B).

If it is from a known node, then the packet is processed normally (step C). If it is from a new node, then the end node controller generates a sequence number inquiry packet and then sends it to the destination from which the packet was received (step D). A sequence number response packet is then received from the new node (step E). The sequence number response packet contains the local node's sequence number assigned to it by the new node. The local node then stores the sequence number for later use (step F).

FIG. 3 is a flowchart illustrating the operation of a sequence number response procedure according to the present invention. This procedure is executed at the end node controller of any node to determine whether a sequence number response packet is to be sent. First, the terminal node controller receives and decodes the packet (step G). The packet is then tested to determine whether it is a sequence number inquiry packet (step H). If it is not, then the end node controller continues normal processing (step J). However, if so, then the end node controller generates and sends out a sequence number response packet bearing its sequence number for the requesting node (step K). The sequence number may be determined by maintaining a counter of the number of sequence number inquiry packets received by the node.

In a preferred embodiment of the invention, the sequence number assigned to the local node by the remote node is used by the end node controller of the local node to cause the local node to flood packets to the remote node via a link common to the two nodes. Help decide whether to send or not. If the sequence number is low, then the local node indicates that it is one of several nodes that can communicate with that remote node and that it has that remote node in its destination address list for flooding packets. Recognize that you will use that knowledge to add nodes. If the sequence number is high, then the local node has many other nodes previous contacts with that remote node, and
It then recognizes that it uses that knowledge to omit that remote node from its list of destination addresses for flooding packets.

FIG. 4 is a flowchart illustrating flooding decisions in accordance with the present invention. First, the end node controller of the local node receives a packet that requires routing (step L). It marks all its links with sequence numbers less than a preselected value intended to indicate a flooding threshold as "still sent" (step M). In this way, links with low sequence numbers are flagged.

The end node controller then tests the link and determines whether the link is in a "still being sent" state (step N). If so, the end node controller causes the packet to be "directed" to the link intended for the remote node and sets the flag marked "still to be sent" as erroneous (step P ). Targeted flooding packets are then sent (step Q
).

The link test (step N) is repeated until all "still to be sent" links are taken into account.

The above procedure is easily implemented in a suitable computer program. For example, a sample of the coding used to make a decision whether to transmit a packet to a particular node is as follows.

＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊＊The following are "routing" and "routing support."

(Left below) #znclude”net, hl’ #xnclude”plaseform,h”#1nc
lude”ut 1s, hflstatxc unsx
gned char conF ENGLE[]-FILEs
taseIc unslgned char conDA
TE[] = DATEstatxc unsx
gnad char conT Engineering ME[]-TIME
;vold Pr1deL3LanWan(1(np
rhntf'(11s ts %s\n", c
onFILEconDATE, conTIrzuE);I
LAYER] DATA 13 dat.

LAYER3DATA ”13 dp(l (r
eturn(&13 dajin)+l/”endof
13 dp(1” */ Ll-MD GeSeNewPxd() (while
e(! ++l] dat, pid);
reseurn13-dp()->pid ;1 /★ Meeting ** Takashi ☆ * ★ Meeting ☆☆ Meeting ☆ Meeting ☆ ROUT
ING★★☆☆Url★★★★＊會★會★★★Sac ☆／／★
called with all ineri
gible N0DEs Dis Qualify
d”/5tat1c N0DE ☆ Setell (numl unslgned 1nt num :rag
xster lnt x :raqxste
r N0DE l+np ;for (l
x O, np = nt 5tart(1;
i < LLlJM N0DES, np++
1++ ) xf(np->1n-the-running 1xf
(np->rcvd 1nfo, seq num <=
numnp->5ent seq num<
w+ num 1return np : r6turn NυLLNP ; ) /★end of 'tell(numl'
”/N0DE *most-qualified(d
ev dest ptr, mood )DE
V ADDR*dev dest ptr ;L3
MOOD The mood following code is used to determine the sequence and number for the node.

LOCAL LITEEXT seq num [n
urn nodeslIJCO(JNT 5equen
ce number (N5UBNET net) FA
ST ARG l; FACT N0DE *np LITEEXT *seq num pt
r.

(From VO i memset(seq num, (UT
EXTllo, NumNoclesl ;for(
i=O, np=nt; i<=Max
Node-1n Use; np++, l+
10) if(ActxveNode(np, net)
1seq-num [np-> 5ent seq nu
m] = 1 :for(seq num ptr
= seq-num+l, l=l;i
= <= Max Node In Us
e + 1; seq num ptr++
, l+ten) if (★seq num-ptr =
= (TINY)O1return l ; /node table is full ”/re
turn NumNodes.

)/★end of 'sequence-number
Although the system has been described to illustrate a preferred embodiment, variations and modifications to the system described herein that are within the scope of this invention are certainly possible. It will therefore occur to those skilled in the art that the foregoing description should therefore be considered merely illustrative and the invention should be limited only in accordance with the claims appended hereto.

[Brief explanation of the drawing]

FIG. 1 is a schematic diagram of a grid of bucket nodes illustrating the propagation of a single packet through a network according to the present invention. FIG. 2 is a flowchart illustrating the operation of sequence number acquisition according to the present invention. FIG. 3 is a flowchart illustrating the operation of sequence number response in accordance with the present invention. FIG. 4 is a flowchart illustrating flooding decisions in accordance with the present invention. 10.12.14.16.18.20.22 in the figure
．． 24.26.28, and 30 are grids of bucket nodes, 32.34.36.38.40.42.
44.46.48.50.52.54.56.58.6
0.62.64.66.68.70.72.74.76
．． 78 and 80 are links.

Claims (8)

[Claims] (1) A method for routing flooding packets to nodes in a packet-switched communication network, the method comprising: assigning a link sequence value at each local node to each communication link with the local node; the link sequence assigned to a common communication link between each said local node and each remote node such that the local node is informed of the link sequence value assigned by each remote node to the communication link with said local node; A method comprising: exchanging values; and transmitting flooding packets only over communication links having a link sequence value with a priority that exceeds a link sequence value threshold. The method of claim 1, further comprising: (2) indicating to the source node that flooding packets may only be rebroadcast to other nodes away from the source node to avoid retransmitting the packets. . 3. The step of directing includes the step of directing the flooding packet to a node located in a far semicircular region bisected by an axis between the source node and the intermediate rebroadcast node. The method described in. (4) In a packet-switched communication network having independent nodes capable of communicating via a communication link, a method for routing flooding packets to nodes in the network, the method comprising: and testing the packet to determine whether the packet originates from a new node, the new node having no established communication link with the local node. a sequence number query from the local node to the new node via the communication link to query the sequence number assigned by the new node to the local node for the communication link; assigning a sequence number at the new node to the communication link established with the new node, the sequence number identifying a relative priority of communication over the communication link; , receiving a sequence number response at the local node from the new node, the sequence number response packet including the sequence number, and storing the sequence number at the local node; and flooding packets at the local node. and the flooding packet is a message intended for all nodes in the packet network, and further the link sequence value has a higher priority than a preselected link sequence value threshold. transmitting the flooding packet again only to the new node. 5. The method of claim 4, wherein the assigning step includes assigning the link sequence for a link to be established. 6. The method of claim 5, wherein the first link established is assigned a lowest link sequence. 5. The method of claim 4, further comprising: (7) indicating that flooding packets may only be rebroadcast away from the source node to avoid re-sending the packets to the source node. (8) said step of instructing that the flooding packets be directed to a node located in a far semicircular region bisected by an axis between the source node and said local node serving as an intermediate rebroadcast node; 8. The method of claim 7, comprising: