WO2013115802A1

WO2013115802A1 - Zig zag routing

Info

Publication number: WO2013115802A1
Application number: PCT/US2012/023371
Authority: WO
Inventors: Sung-Ju Lee; Jung Gun Lee; John S. Balian; Raul Hernan Etkin; Richard S. Davis; Scott A. Lindsay; Tom Hogan
Original assignee: Hewlett-Packard Development Company, L.P.
Priority date: 2012-01-31
Filing date: 2012-01-31
Publication date: 2013-08-08

Abstract

Messages in a wireless multi-hop network comprising two or more rows of nodes are routed from a source node to a destination node along a path which zig-zags between a first row of nodes and a second row of nodes.

Description

ZIG ZAG ROUTING

BACKGROUND OF THE INVENTION

[0001] Many wireless networks comprise a plurality of nodes and communicate messages from a source to destination via a multi-hop path. With multi-hop routing, each node forwards a message to a next hop on the path towards its destination and eventually, usually after several hops, the destination node receives the message.

[0002] As wireless networks use a shared communication medium, e.g. the air, interference can be an issue. One way to address this is for different nodes to use different channels, such that communications on the two different channels do not interfere with each other.

BRIEF DESCRIPTION OF THE DRAWINGS

[0003] Some examples are described in the following figures:

[0004] Figure 1 shows an example of a wireless network in which nodes all use the same wireless channel for communication;

[0005] Figure 2 shows an example of a wireless network in which adjacent nodes use different wireless channels for communication;

[0006] Figure 3 shows an example of a wireless network in which the nodes route messages via a zig-zag path;

[0007] Figure 4 shows another example of a wireless network in which the nodes route messages via a zig-zag path; and

[0008] Figure 5 shows an example structure of a node for use in a multi-hop wireless network.

DETAILED DESCRIPTION

[0009] There are many types of wireless network, for example WLAN 802.11x type networks, cellular networks, WiMax and 802.15.4 ZigBee etc.

[0010] A wireless multi-hop network is a wireless network in which messages are forwarded from a source to a destination along a path comprising one or more intermediate nodes. Each node determines which node is the next in the path towards a message's destination. The message is then forwarded to this 'next hop'. For example, each node may use a routing table to decide which next hop to forward a message to. In another approach, known as "source routing", the source node may specify a full list of nodes in the path to the destination and place this list in the packet header; then each node forwards the packet to the specified next hop without having to make a decision. Still another approach is pure level 2 bridging in which nodes learn about other nodes based on MAC addresses and interfaces.

[0011] In the context of this disclosure, the 'next hop' may be an intermediate node or the destination node itself.

[0012] In many network topologies there may be two or more rows of nodes and each row of nodes forwards messages along a multi-hop path to a head node such as a sink node or gateway node.

[0013] A gateway node is a node which connects the wireless network to the internet. A sink node is a node which acts as a centralized decision making unit (e.g. in a wireless sensor network one node may be designated as a sink node for collecting data and/or processing data and/or making it available to an end user). Communication may be two-way so that the sink or gateway node may act as the destination node when receiving messages from other nodes in the network and act as the source node when sending messages to other nodes in the network.

[0014] Two examples of wireless networks are a wireless sensor network and a wireless network for accessing the internet. A wireless sensor network (WSN) is a network comprising a plurality of spatially distributed sensor nodes which may monitor physical or environmental conditions such as the temperature, sound, vibration, pressure, motion, light levels or pollution etc. Typically nodes in a wireless sensor network pass their data through the network along a multi-hop path to or from a sink node. While sensor node networks with only one-way communication exist, two-way communication facilitates remote control of the sensor nodes. Modern wireless sensor networks sometimes have several sink nodes or a distributed control system in which any node may become a sink node. The techniques in this disclosure may be applied to one-way or two-way wireless sensor networks and to those with a central hierarchy or with a distributed control structure.

[0015] A wireless network for accessing the internet comprises a plurality of nodes which are wireless access points and/or wireless repeaters and at least one node which is a gateway. The nodes in the wireless network forward messages along a multi-hop path to and from the gateway to allow anyone connecting to an access point, with appropriate permission, to access the internet. Such networks are increasingly used in developing countries which do not yet have sufficient infrastructure to allow widespread wired internet access.

[0016] Figure 1 shows a network topology for a wireless network comprising two rows of nodes. Nodes A to E in the first row form a multi-hop path to a head node K, which may for example be a gateway or a sink node. Thus, for example, a message may be forwarded from node A to node K via nodes B, C, D and E. Node B and other nodes may also forward their own data to node K along this path. Likewise nodes F to J, in the second row of nodes, form a multi- hop path to the head node K.

[0017] At least some of the nodes will have their own traffic to send to the gateway or sink node K. In addition each node will receive and relay (forward) traffic from neighboring nodes which are farther away from the sink or gateway node. Thus, where node A has traffic to send to node K, this traffic will be forwarded to node B, which will forward the traffic to node C, etc until node K is reached. [0018] In this example all the nodes in the network use the same channel (e.g. the same carrier frequency). So, for example, node A may send data to node B. After node B receives the data from node A, node B forwards the data to node C. However, node A can also receive, or Overhear', node B's transmission to node C. Since the nodes are operating on the same channel, while node B transmits, node A cannot transmit at the same time. The problem is even worse if node C's transmission can be overheard by both nodes A and B. Furthermore, nodes in the first row may also interfere with nodes in the second row. Various schemes such as Carrier Sense Multiple Access/Collision Avoidance (CSMA/CA) exist in order to prevent or mitigate collision of signals from different sources on the same channel. However, the network throughput can be significantly reduced, as it not possible for nodes which may interfere with each other to transmit at the same time.

[0019] Directional antennas and/or beamforming may be used to reduce spatial interference with adjacent rows, e.g. node B may have a first directional antenna associated with its first radio and pointed at node A, and a second directional antenna associated with its second radio and pointed at node C. However, even then, if the second row is close enough, or the signal strong enough, signals from the second row may still interfere with the first row. Highly directional antennas also increase the likelihood and range of 'boresight' interference; i.e. nodes in the same 'line of sight' or on the same row, receiving signals which are not intended for them, e.g. node C overhearing the transmissions of node A.

[0020] The distance by which nodes are separated and the signal strength both affect the degree of interference and the number of nodes which may interfere with each other.

[0021] Figure 2 shows another example of a wireless network. In this example, adjacent nodes use different channels in order to reduce or prevent interference. Specifically the nodes have multiple radios and use a different channel for each radio. For example, the different channels may be orthogonal to each other or a filter may be used to ensure each radio receives the desired channel (possible implementations include Orthogonal Frequency Division Multiplexing, or Frequency Division Multiplexing with a filter for each side band).

[0022] In Figure 2, node A uses a first channel indicated by solid lines to communicate with node B. Both node A and node B each have a radio tuned to the first channel. Meanwhile node B and node C communicate with each other on a second channel (indicated by a dotted line). Thus node B has a first radio tuned to the first channel for communication with node A and a second radio tuned to the second channel for communication with node C.

[0023] Further, node C uses a third channel (indicated by a dashed line) to communicate with node D. This is because otherwise, in this example, the transmissions of node C to node D might be overhead by node A (this is especially the case were node A uses a directional antenna - this is known as boresight interference). Thus, node A cannot use any of the same channels as node C. Of course the probability of interference depends upon the strength of the signal, distance between the nodes and other factors. In this example node As transmission cannot reach node D, as it is too far away. Hence node D can use the first channel to communicate with node E and node E uses the second channel to communicate with node K.

[0024] Note that nodes in the second row perform a similar task. However, they use different channels to avoid interference with the first row of nodes. Thus, node F uses a fourth channel (indicated by a dotted and dashed line) to communicate with node G, node G uses a fifth channel (indicated by a bold dashed line) to communicate with node H, and node H uses a sixth channel (indicated by a bold dotted line) to communicate with node I. Node I then uses the fourth channel to communicate with node J and node J uses the fifth channel to communicate with node K. Thus in the example of Figure 2, a total of six different channels are required to mitigate interference.

[0025] Figure 3 shows an example which uses a zig zag routing topology. As in the previous examples there are two rows of wireless nodes and a head node K. In some implementations there may be other nodes or more than two rows of nodes, but for clarity only two rows are shown in Figure 3. [0026] The first row of nodes comprises a first node A, third node B, fifth node C, seventh node D and ninth node E. The second row of nodes comprises a second node F, fourth node G, sixth node H, eighth node I and tenth node J. The head node K may for example be a gateway or a sink node. For the purposes of this disclosure the head node (e.g. gateway or sink node) counts as being on both the first and second rows, i.e. it may be the next hop on a zig zag path from a node in either the first or second row.

[0027] Messages from a source node to a destination node are routed along a multi-hop path which zig-zags between the first row of nodes and the second row of nodes. For example if node A has a message to transmit to node K, then node A sends the message to node G as the next hop node. Typically the message will be sent in one or more packets having a destination address in the header or elsewhere indicating its final destination and the receiving node will read the destination address before deciding where to route the incoming packet. The forwarding decision may be made at layer 2 or layer 3. Typically a forwarding decision will be made by each node on the path (e.g. with reference to a routing table). Alternatively the entire zag-zag path may be predetermined by the source node if a "source routing" protocol is used. Still another routing approach is pure level 2 bridging in which each node makes a forwarding decision, but in which nodes learn about other nodes based on MAC addresses and interfaces rather than by reference to a routing table. In the context of this disclosure and the appended claims 'routing' should be interpreted to include pure level 2 bridging as well as other methods such as source routing, or routing decision at each node by reference to a routing table or a layer 3 routing protocol etc.

[0028] Node G forwards the message to node C, which forwards the message to node I which forwards the message to node E, which forwards the message to node K. The same zig zag routing path may be used to route messages between node K and node A (i.e. the reverse of the above), or to route messages between node K and other nodes in the first row (e.g. between nodes C and K), or between certain nodes in the first row (e.g. between nodes A and C), or between certain nodes on the first row and certain nodes on the second row (e.g. between node A and node I).

[0029] Likewise, a second zig zag path is used to route messages between node F on the second row and the head node K. Thus, node F sends messages to node K via a second zig zag path (as shown in Figure 3), which traverses nodes F, B, H, D, J and K and which alternates between the first and second rows at each hop. The second zig-zag path may be used to route messages between other nodes on the second row and node K (or vice versa) or to route messages between certain nodes on the second row or to route messages between certain nodes on the second row and certain nodes on the first row.

[0030] Consider the path from node A to node G to node C, in particular where the communication between the nodes is carried out by directional antennas or beamforming antennas. As the communication from node A to node G does not point in the same direction as the next hop (node C), the problem of boresight interference is mitigated. Node C cannot hear the traffic between node A and node G. Thus, whereas six different channels were needed to mitigate interference in the example of Figure 2, in the example of Figure 3 only four different channels are needed.

[0031] Thus in Figure 3, a first channel (shown by a solid line) is used to communicate between nodes A and G. Meanwhile node G communicates with node C by a second channel (shown by a bold dashed line). As node C cannot overhear the first channel traffic between nodes A and G, node C is able to communicate with node I by (re-)using the first channel. Similarly, node I communicates with node E using the second channel and node E communicates with node K using the first channel.

[0032] Nodes F and B may overhear the traffic between nodes A and G, so node F communicates with node B using a third channel (shown by a dotted and dashed line). Meanwhile node B communicates with node H by using a fourth channel (shown by a dotted line) - and in this way it does not interfere with the traffic between nodes G and C. Node H communicates with node D using the third channel and node D communicates with node J using the fourth channel. Finally node J communicates with node K using the third channel. [0033] Thus the example of Figure 3 only requires four different channels. Thus, in this particular example, for each zig zag path two channels are used and at each hop on a zig zag path the channel used alternates between these two channels.

[0034] Use of multiple channels mitigates interference which might otherwise reduce the throughput of the network. Further, as in the example of Figure 3 only four different channels are used, other channels are left available for other purposes. In some cases, only a limited number of channels may be available, due to the presence of background interference or traffic from other parts of the network, or from other nearby networks. Thus the zig zag routing scheme of Figure 3 may be employed in a wider range of circumstances as it requires fewer channels.

[0035] In a wireless sensor network the sink node may sometimes pull sensor node data from all the nodes in a particular row of the network. That is each node in the row sends data to the sink node. This may be a data sample (sometimes referred to as a 'data trace') for the purposes of checking that the sensor data is being collected correctly. With the (linear) routing topology of Figure 2, the load increases at each node as the path gets closer to the sink node. Thus, in Figure 2, node A sends its own data to node B as the next hop; node B sends both its own data and the data received from node A to node C as the next hop etc. Finally node E sends data from all of nodes A, B, C, D and E to node K. Thus the load on node E is comparatively high.

[0036] In the zig zag routing of Figure 3, the relay load is comparatively reduced. That is, the load is split between the first and second rows. For example, when the sink node K pulls a data trace from nodes in the first row; node A sends its data to node G as the next hop, while node G simply forwards the received data to node C, without necessarily adding its own sensor data. Node C then forwards its own data and the data from node A (received via node G) to node I. Thus, at the end of the row, node E forwards data to node K from nodes A, C and E; while node J forwards data to node K from nodes B and D. The relay load on any particular node is thus reduced compared to the routing paths in Figure 2. [0037] In the examples above, the phrase "comprising two rows of nodes" or "comprising a first row and second row of nodes" should be interpreted as nonexclusive meaning that other elements may be present. For example, the network may have only two rows of nodes, or there may be three or more rows of nodes and there may be other nodes not belonging to a particular row. The techniques described above may be applied to networks with more than two rows of nodes. For example, if there are four rows of nodes, then the first and second rows may use zig zag paths as shown in Figure 3, and nodes in the third and fourth rows may use similar routes which zig-zag between the third and fourth rows. The third and fourth rows may use different channels to the first and second rows if they are close enough for their signals to be overheard by the first and second rows.

[0038] Usually there will be an even number of rows in the network. However, if there is an odd number of rows, e.g. three rows, then nodes in a third row may route messages to nodes in the second row and onto the zig-zag path between the first and second rows until the destination is reached. An example is shown in Figure 4. Node M of the third row communicates with node G of the second row on a fifth channel (shown by a dashed line) and forwards messages there as the next hop to head node K. Node N of the third row communicates with node H of the second row on a sixth channel (shown by a bold dotted line) and forwards messages there as the next hop to head node K. Likewise nodes O and P communicate with respective nodes of the second row on the fifth and sixth channels and node Q communicates directly with head node K on the fifth channel. Thus messages between nodes in the third row and the head node K can be transmitted back and forth along the zig zag path between the first and second rows of nodes. In other cases a zig-zag path between the second and third rows may be used.

[0039] Figure 5 shows an example structure of a wireless node 100 for use in a network such as the networks of Figures 3 or 4. The wireless node 100 comprises a first directional antenna 110 connected to a first radio 115 and a second directional antenna 120 connected to a second radio125. The first and second radios are tunable to two different channels. E.g. the channels may be orthogonal to each other or one or more filters may be used so that each radio only receives the desired channels. The first directional antenna 110 may be pointed at a wireless node in a different row of nodes to the node 100. The second directional antenna 120 may be pointed at another wireless node in a different row of nodes to the node 100. Alternatively the wireless node may have functionality to automatically direct each antenna towards a detected signal on a particular frequency.

[0040] In the above example the first and second antennas are directional antennas, however in other examples the antennas may be 'beamforming' antennas or 'beamforming' antenna arrays capable of concentrating the radio signals in a particular direction. It should also be noted that in some cases the node may have one or more further radios and antennas (e.g. a third radio and associated antenna) for communicating with nodes in other parts of the network outside of the first and second rows.

[0041] Each radio is able to communicate with a processor 130, e .g. the processor may be a CPU, processing unit, ASIC, logic unit, or programmable gate array etc. The processor 130 receives data from and sends data to the first and second radios for wireless transmission. The processor 130 may access and execute machine readable instructions stored on a memory 140; e.g. the memory may be a RAM, ROM, or flash memory etc. Further while a single memory is described here, the contents may be distributed over several separate memories. In this example the memory 140 stores several modules of machine readable instructions 150, 160.

There is a module 150 of machine readable instructions to handle construction of data packets and sending or receiving wireless messages, e.g. in accordance with an 802.11 or other wireless networking standard. There may also be modules for handling other basic operations in the wireless network.

[0042] A module of machine readable instructions 160 handles routing decisions, such as the next hop to which a received message should be sent or the next hop to which a message generated by the wireless node 100 should be sent. The routing decision may make reference to entries in a routing table 170 stored in the memory. The decision may be simple, e.g. to forward all messages received on a first channel used by the first radio out of a second channel used by the second radio. In other implementations the decision may be based on the destination address of a particular packet and an entry in a routing table giving the next hop for a particular destination address. To give a specific example, if a routing decision is made by each node on the path based on a message's destination address, then the contents of the routing table for node C in Figure 3 may be as follows:

[0043] The routing table thus indicates that if node C generates or receives a message having a destination of node K then it should be forwarded to node I as the next hop. If node C receives or generates a message having a destination of node A then it should be forwarded to node G as the next hop.

[0044] In still other implementations the entire zig-zag route may be predetermined by the source node and e.g. included in a header of the packet. In yet other implementations the routing may be by layer 2 bridging in which case each node will learn about the adjacent nodes to which packets should be routed through MAC address learning.

[0045] For the purposes of this disclosure, the basis on which the routing decision is made by the processor (whether using routing rules, reference to a routing table, machine readable instructions, or a combination thereof) may be referred to as 'routing logic'.

[0046] The contents of the routing table and/or the rules used by the routing module 170 may for example be set up by a network engineer when the network is first set up, or later by a system administrator.

[0047] Those skilled in the art will understand that the various modules in the aforesaid example can be combined into one module or further divided into a plurality of sub-modules.

[0048] The zig zag routing described above can be implemented in a wireless network by appropriate configuration and set-up of the network. For example, if the nodes have directional or beamforming antennas, then the antennas may be pointed at nodes in another row rather than nodes in the same row. Further the processor of each node may be provided with routing logic for routing traffic according to the zig-zag paths described above with reference to Figures 3 and 4.

[0049] Each node may be given specific channels for each radio or programmed to select channels in accordance with the channel assignment discussed above. For example, four different channels may be chosen for the network and assigned to radios of the individual nodes accordingly. One example of a rule for assigning channels could be that the radios of an individual node should use different channels and the channels should also be different to those used by the adjacent nodes on the same row. The channels may also be different to the channels used by the nearest node on an adjacent row.

[0050] The above examples with regard to operation of the wireless nodes and routing of messages can be implemented by hardware, software or firmware or a combination thereof. For example the various methods, processes and functional modules described herein may be implemented by a processor (the term processor is to be interpreted broadly to include a CPU, processing unit, ASIC, logic unit, or programmable gate array etc.). The processes, methods and functional modules may all be performed by a single processor as shown in Figure 5 or split between several processers; reference in this disclosure or the claims to a 'processor' should thus be interpreted to mean One or more processors'. The processes, methods and functional modules be implemented as machine readable instructions executable by one or more processors, hardware logic circuitry of the one or more processors or a combination thereof. Further the teachings herein may be implemented in the form of a software product. The computer software product is stored in a storage medium and comprises a plurality of machine readable instructions for making a computer device (which can be a personal computer, a server or a network device such as a router, switch, access point or sensor node, etc.) implement the method recited in the examples of the present disclosure.

Claims

WHAT IS CLAIMED IS:

1. A method of routing messages in a wireless multi-hop network comprising two or more rows of nodes; the method comprising routing messages from a source node to a destination node along a path which zigzags between a first row of nodes and a second row of nodes.

2. The method of claim 1 wherein the destination node is a gateway or sink node.

3. The method of claim 1 wherein the source node is a gateway or sink node.

4. The method of claim 1 wherein at each hop the zig zag path crosses from the first row of nodes to the second row of nodes or from the second row of nodes to the first row of nodes.

5. The method of claim 1 wherein the network is a sensor node network.

6. The method of claim 1 wherein the nodes are wireless access points or wireless repeaters or a gateway.

7. The method of claim 5 wherein a sink node pulls data from sensor nodes in the first row of nodes via two zig zag paths, each zig zag path alternating between the first row and second row of nodes at each hop, nodes on the second row of nodes forwarding received data to a node on the first row of nodes, nodes on the first row of nodes adding their own data to the message and forwarding to a node on the second row of nodes.

8. The method of claim 1 wherein at each hop along the zig zag path, the channel on which a message is forwarded between nodes alternates between two different channels.

9. A wireless node for use in a multi-hop wireless network comprising a first row of nodes and a second row of nodes; the wireless node comprising:- two or more radios each with an associated antenna; a processor and a memory storing machine readable instructions executable by the processor to route a message towards a destination node by sending the message to a next hop node which is on a different row to the row on which the wireless node is located.

10. The wireless node of claim 9 wherein a first radio of the node is tuned to a first channel and the second radio is tuned to a second channel.

11. A multi-hop wireless network comprising a first row of nodes including a first node and a third node and a second row of nodes including a second node and a fourth node;

each node having two or more radios and two or more antennas;

wherein the first node is to route messages to the fourth node;

wherein the second node is to route messages to the third node;

wherein the fourth node is to route messages to a next hop node on the first row of nodes;

wherein the third node is to route messages to a next hop on the second row of nodes.

12. The network of claim 11 wherein

the third node is to route messages to a sixth node in the second row;

the fourth node is to route messages to a fifth node in the first row;

a first wireless channel is used for communication between the first node and the fourth node and between the fifth node and its next hop node;

a third wireless channel is used to communicate between the second node and the third node and between the sixth node and its next hop node; a second wireless channel is used to communicate between the fourth node and the fifth node;

a fourth wireless channel is used to communicate between the third node and the sixth node.

13. The network of claim 11 wherein the network is a sensor node network.

14. The network of claim 13 wherein the network includes a sink node which is to pull data from nodes in the first row; whereby nodes in the second row are to forward packets to a next hop node, while nodes in the first row are to add their own data and then forward the packet to a next hop node.

15. The network of claim 11 wherein the nodes are wireless access points or wireless repeaters and the destination node or the source node is a gateway.