KR101040556B1 - The Wireless Mesh Network of Cross-Layer Optimization and Implementation Method Thereof - Google Patents

Abstract

Disclosed are a cross-layer optimized wireless mesh network and an implementation method thereof, which can achieve optimization and fairness without changing an existing protocol (UDP / TCP) applied to a user's wireless device.

The disclosed wireless mesh network has a structure of a wireless mesh network to a user's wireless device and the Internet, a wireless access network that is a network to an ingress mesh access point for transmitting and receiving packets wirelessly with a user's wireless device, and a connection with the Internet. For the ideal solution in a wireless backhaul network, a wired network network is connected to the source protocol and the layered crossover optimization protocol of the radio access network, and the wireless backhaul network connects the link protocol and the layered crossover optimization protocol of the wired network. By using both the inferred link algorithm and the source algorithm, an optimized network is constructed.

Wireless mesh network, wireless backhaul network, layered crossover optimization algorithm,

Description

The Wireless Mesh Network of Cross-Layer Optimization and Implementation Method Thereof}

The present invention relates to the implementation of a cross-layer optimized wireless mesh network, and more particularly, to a wireless multi-hop network without changing the existing protocol (UDP / TCP) applied to a user's wireless device. The present invention relates to a layered optimized wireless mesh network and an implementation method thereof to achieve optimization and fairness.

In general, the wireless mesh network technology is a technology that supports Internet access by multi hop instead of single hop based on the existing wireless LAN (IEEE 802.11 standard) technology.

Supporting data networks in multi-hop wireless networks, unlike single-hop wireless and wired networks, not only makes resource allocation more complex, but also limits the resources. In the case of wired networks, the problem of resource allocation can be easily solved by separating the network connection so that resource sharing does not occur. In addition, in the case of a single hop wireless network, the resource allocation problem and resource limitation are overcome by appropriately distributing the resources shared by the centralized station.

However, the resource allocation problem and the resource limitation problem in the wireless multi-hop network have a distinct difference. The reason is that resources must be divided not only including neighboring nodes that want to send and receive data due to hidden node problems that the invisible node affects, but also invisible nodes that are neighbors of the node. This resource sharing can affect the entire network, such as the neighbors of the neighborhood, and the neighbors of the neighborhood. In addition, unlike a conventional multi-hop network, MANET (Mobile Ad-hoc Network), the wireless mesh network has a characteristic that data flow characteristics occur between a gateway and a node, a node, and a gateway. There is a drawback to being prone to congestion.

In addition, as a way of efficiently allocating limited resources, wireless mesh networks must solve problems in a distributed manner rather than structurally centralized. This is a very complex and difficult problem, but fortunately, there is a known way to find an optimized solution.

The method for finding the optimized solution is to find a cross-layer optimization solution using a mathematical model, as shown in Table 1 below.

ode-Centric Algorithms Link-Centric Algorithms Flow Control at Source

Scheduling at Link

Each definition of the algorithm of Table 1 is as follows.

f: flow index

d: index of queue created by destination

l: Index of queue created by link

:플로우 f의 전송 레이트)x: data rate (e.g.

Transfer rate of flow f)

: 목적지가 d인 큐의 길이,

: 링크가 l인 큐의 길이,

: 링크 l에서 목적지가 d인 큐의 길이)q: the length of the queue, for example

: Length of queue at destination d,

: Length of queue with link l,

: Length of queue at destination l at destination d)

: 링크 l의 전송 용량c: capacity (

: Transmission capacity of link l

D: Destination set of flows (D (l): Destination set of flows on link l

L: set of links (L (f): set of links through which flow f passes)

U: Utility function

Hierarchical optimization algorithm can be largely divided into Node-Centric algorithm described based on node (mesh node) and Link-Centric algorithm described based on link (wireless link between mesh nodes).

Both of these algorithms have a Source Algorithm that determines how much of each flow flows and a Link Algorithm that determines which link should be used.

Briefly, this algorithm is similar to the principle of water flowing in nature. Determining the amount of data that a node needs to send to the network and the link that is the resource that should be used for transmission is all determined by looking at the length of the queue represented by q. As water flows, the priority of the water to flow depends on its height, and if the height of the water is high, the path through which the water flows is congested.

Looking closely at the operation of the Node-Centric and Link-Centric algorithms, the principle is the same, but the algorithm works differently.

로 표현된 값은 자신 노드의 목적지 큐 길이에서 이웃한 노드의 목적지가 같은 큐의 길이를 뺀 값 중 해당 링크 l에서 제일 큰 값을 의미한다. 링크 알고리즘은 이값과 해당 링크 l의 링크 용량의 곱이 네트워크 전체에 걸쳐 최대가 되도록 링크를 동작시켜야 하며, 소스 알고리즘은 각 목적지 별로 네트워크 큐의 길이의 효용함수를 적용한 전송속도로 전송을 해야한다.In the Node-Centric algorithm, all nodes manage queues by destination.

The value represented by means the largest value in the link l among the subtraction of the destination queue length of its own node minus the length of the same queue. The link algorithm must operate the link so that the product of this value and the link capacity of the corresponding link l is maximum throughout the network, and the source algorithm should transmit at the transmission rate applying the utility function of the network queue length to each destination.

 All nodes in the Link-Centric algorithm manage queues on a per link basis. Here, the link algorithm should operate the link so that the link queue length and the capacity of the link are maximized throughout the network, and the source algorithm transmits the utility function to the sum of the link queue lengths in all the flows. Determine the speed.

The conventional hierarchical crossover optimization algorithm described above is known as an algorithm for achieving network optimization in consideration of fairness by a utility function determined according to a user's request in a multihop wireless network. Of course, decentralized algorithms that achieve this optimization are also known. However, there are many difficulties in applying it to a real network.

Among them, the hierarchical cross-optimization algorithm is provided with a link algorithm performed at the link and a source algorithm executed at the source. The link algorithm can be performed at a node supporting a network (wireless mesh router), so it is not difficult to apply to a network device. However, in the case of the source algorithm, it must be performed on individual wireless devices, so the protocol of the user's wireless device must be modified to achieve optimization. This means that standardized protocols other than the existing UDP / TCP protocols are standardized and adopted in user equipment, which can be very difficult and cumbersome.

Therefore, the present invention is proposed to apply the layered optimization algorithm to the real network, so that even if the user's wireless device performs the current protocol (UDP / TCP), it has the same result as performing the layered crossover optimization algorithm. As,

An object of the present invention is to provide a hierarchically optimized wireless mesh network and an implementation method thereof to achieve optimization and fairness without changing an existing protocol (UDP / TCP) applied to a user's wireless device.

The layer-optimized wireless mesh network according to the present invention for solving the above problems,

The wireless mesh network structure of the user's wireless device and the Internet, the wireless access network which is a network to the ingress mesh access point for transmitting and receiving packets wirelessly with the user's wireless device, the wired network network for connection to the Internet, The source protocol and the layered crossover optimization protocol of the radio access network are matched, and the wired network network is separated into a wireless backhaul network matching the layered crossover optimization protocol.

In addition, the wireless backhaul network,

Has an Ingress mesh router that analyzes the headers of packets, manages queues, and allocates bandwidth,

After separating the queue into an elastic queue and a non-elastic queue in the ingress mesh router, the packet header is analyzed to separate the TCP data from the non-TCP data. And store the non-TCP data in the non-elastic queue.

In addition, the ingress mesh router,

A link for controlling the non-elastic queue to be serviced before the elastic queue, wherein the non-elastic queue allocates a static bandwidth and the elastic queue determines which link to use Dynamic bandwidth is allocated according to the algorithm.

A layer-optimized wireless mesh network implementation method according to the present invention for solving the above problems,

The structure of the wireless mesh network to the user's wireless device and the Internet is divided into a wireless access network, a wireless backhaul network, and a wired network. The divided wireless backhaul network allocates resources using a layered crossover protocol applying an ideal solution. The radio access network and the wired network network receive an optimized service using an existing standard protocol (TCP / UDP) in consideration of an internet connection and a connection with a user.

The wireless backhaul network includes an ingress mesh router that analyzes the header of the packet and manages the queue, wherein the ingress mesh router allocates a queue to an elastic queue and a non-elastic queue. After separating into a non-elastic queue, packet headers are analyzed to separate TCP data and non-TCP data so that TCP data is stored in the elastic queue, and the non-TCP data is stored in the non-elastic queue. It is characterized by.

In addition, the wireless backhaul network,

The elastic queue may be implemented as a per-destination queue or a per-link queue according to an implementation of a node-centric algorithm or a link-centric algorithm to perform load sharing.

In addition, the wireless backhaul network,

A link for controlling the non-elastic queue to be serviced before the elastic queue, wherein the non-elastic queue allocates a static bandwidth and the elastic queue determines which link to use Dynamic bandwidth is allocated according to the algorithm.

In addition, the wireless backhaul network,

In order to connect the existing source algorithm (TCP) and the hierarchical cross optimization algorithm, TCP data is first stored in a TCP adaptation queue before being stored in the elastic queue.

In addition, the wireless backhaul network,

When transmitting data from the TCP adaptation queue to the elastic queue, a response is randomly transmitted according to the service rate to control the source algorithm of the hierarchical optimization algorithm to be the same as the TCP rate.

According to the present invention, there is an advantage in that optimization and fairness can be achieved without modifying a protocol of a user in a wireless mesh network.

Hereinafter, described in detail with reference to the accompanying drawings a preferred embodiment of the present invention. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail.

FIG. 1 is an overall structure diagram of a wireless mesh network to which the present invention is applied, and includes a wireless access network 510 and an ingress mesh access point 200 which are networks for transmitting packets wirelessly with a user wireless device 100. A wireless backhaul network 520 that performs packet transmission with the user wireless device 100 and transmits a packet through a wired network with the Internet 400 through a mesh gateway 300, the mesh gateway 300 and the Internet ( Located between 400 and includes a wired network 530 for transmitting packets over a wired network.

Here, the ingress mesh access point 200 and the mesh gateway 300 respectively analyze packet headers, manage queues for storing packets to be transmitted, and include ingress mesh routers 210 and 310 for allocating resources. do.

4A and 4B are flowcharts illustrating a method for implementing a layer-optimized wireless mesh network according to the present invention. FIG. 4A is a flowchart illustrating a static network determination method, and FIG. 4B is a flowchart illustrating a dynamic network determination method.

As shown in FIGS. 4A and 4B, the hierarchically optimized wireless mesh network implementation method according to the present invention includes: dividing a wireless mesh network structure into a wireless access network, a wireless backhaul network, and a wired network (S101). Applying a layered crossover optimization protocol to the classified wireless backhaul network, and applying the existing standard protocol (TCP / UDP) to the wireless access network and the wired network (S103), the ingress mesh router of the wireless backhaul network Separating the queue into an elastic queue and a non-elastic queue in step S105, analyzing a header of a packet to be transmitted by the ingress mesh router in step S107, and in the case of TCP data as a result of the packet header analysis, a TCP adaptation queue ( Primary storage in the TCP adaptation queue, and secondary storage of the TCP data stored in the TCP adaptation queue Steps (S111 ~ S113) and; When storing the TCP data stored in the TCP adaptation queue in the elastic queue, transmitting a random response (Fake-ACK) to the packet input side according to the service rate (S115), when the packet header analysis result is not TCP data Storing a packet to be transmitted in a non-elastic queue (S117), allocating a bandwidth after storing the packet to be transmitted (S119), first serving (transmitting) a packet stored in the non-elastic queue, And next serving the packet stored in the elastic queue (S121).

A method of implementing a layer-optimized wireless mesh network according to the present invention as described above will be described in detail with reference to FIGS. 1 to 4.

First, in the present invention, as shown in FIG. 1, the entire wireless mesh network structure is divided into a radio access network 510, a wireless backhaul network 520, and a wired network network 530 (S101). That is, the interval between the user wireless device 100 and the ingress mesh access point 200 for transmitting a packet wirelessly is divided into a wireless access network 510, and the ingress mesh access point 200 and the mesh gateway 300 are provided. The interval between and is divided into a wireless backhaul network 520, and the interval between the mesh gateway 300 and the Internet 400 for transmitting packets by wire is divided into a wired network 530.

Since the wireless backhaul network 520 is a network designed and provided by a network provider, a layered crossover optimization protocol applying an ideal solution is applied, and the other two networks (wireless access network and wired network network) consider an internet connection and a user connection. By applying a protocol (UDP / TCP) according to the existing standard (S103).

In other words, the user using the wireless mesh network can use the existing protocol as it is and receive the optimized service provided by the wireless backhaul network. To this end, the matching of the existing protocol and the layered crossover optimization protocol should be performed normally. A method of controlling bandwidth allocation as a resource is used.

That is, the ingress mesh router 210 serves to analyze headers of packets, manage queues, and allocate bandwidth, as shown in FIG. 2 in the ingress mesh router 210. Likewise, the queue is divided into an elastic queue and a non-elastic queue (S105).

After analyzing the header of the packet to be transmitted (S107), the TCP data and the non-TCP data is separated, and the separated TCP data is stored in the elastic queue (S111), the non-TCP data is the non-astic queue The data is stored in S117.

In this case, the elastic queue is implemented as a per-destination queue or a per-link queue according to the implementation of the node-centric algorithm or the link-centric algorithm to perform load sharing.

Ingress mesh router 210 may also allow the non-elastic queue to have a higher priority than the elastic queue so that the non-elastic queue is serviced before the elastic queue, i.e. The stored packet is controlled to be transmitted first.

On the other hand, the connection between the existing source algorithm (TCP) and the hierarchical cross optimization algorithm is implemented using a fake-ACK as shown in FIG.

That is, TCP data is first stored in a TCP adaptation queue before being stored in the elastic queue, and the transmission from the TCP adaptation queue to the elastic queue is serviced by the source algorithm of the hierarchical crossover optimization algorithm (S113). ).

At this time, when the data is transmitted from the TCP adaptation queue to the elastic queue, a fake response is transmitted according to the service rate to control the source algorithm of the hierarchical crossover optimization algorithm and the TCP rate to be the same. (S115).

After that, the bandwidth is allocated, the non-elastic queue is serviced first, and then the elastic queue is serviced (S119 to S121). Here, the non-elastic queue allocates a static bandwidth (1/10 to 1/3 of the available bandwidth), and the elastic queue is a dynamic bandwidth according to a link algorithm. ) Will be assigned.

According to the present invention as described above, the wireless backhaul network is able to build an optimized network using both the link algorithm and the source algorithm inferred by the ideal solution. In the case of UDP, a user using the network is used within a certain bandwidth by a non-elastic queue, and in the case of a TCP user, the service is applied at a rate suggested by an optimization algorithm.

The present invention is not limited to the above-described specific preferred embodiments, and various modifications can be made by any person having ordinary skill in the art without departing from the gist of the present invention claimed in the claims. Of course, such changes will fall within the scope of the claims.

1 is an overall structural diagram of a layer optimized wireless mesh network according to the present invention.

2 is a structural diagram of a queue according to data type in the present invention.

3 is a conceptual diagram of storing data in the elastic queue of FIG.

4A is a flowchart illustrating a static network determination process in a method for implementing a layer-optimized wireless mesh network according to the present invention.

4B is a flowchart illustrating a dynamic network determination process in a layer optimized wireless mesh network implementation method according to the present invention.

<Explanation of symbols for main parts of the drawings>

100... User wireless device

200... Ingress Mesh Access Point

210... Ingress mesh router

300... Mesh gateway

400... Internet

510... Wireless access network

520... Wireless backhaul network

530... Wired network

Claims (9)

In a wireless mesh network, The wireless mesh network structure of the user's wireless device and the Internet, the wireless access network which is a network to the ingress mesh access point for transmitting and receiving packets wirelessly with the user's wireless device, the wired network network for connection to the Internet, A layer-optimized wireless mesh network matching source protocols of a radio access network and a hierarchical crossover optimization protocol, and separated into a wireless backhaul network matching the wireline network protocol with the hierarchical crossover optimization protocol. The wireless backhaul network of claim 1, wherein: Has an Ingress mesh router that analyzes the headers of packets, manages queues, and allocates bandwidth, After separating the queue into an elastic queue and a non-elastic queue in the ingress mesh router, the packet header is analyzed to separate the TCP data from the non-TCP data. And store the non-TCP data in the non-elastomeric queue. The method of claim 2, wherein the ingress mesh router, A link for controlling the non-elastic queue to be serviced before the elastic queue, wherein the non-elastic queue allocates a static bandwidth and the elastic queue determines which link to use A layer optimized wireless mesh network comprising dynamic bandwidth allocation in accordance with an algorithm. In the method of implementing a wireless mesh network, The structure of the wireless mesh network to the user's wireless device and the Internet is divided into a radio access network, a wireless backhaul network, and a wired network, wherein the divided wireless backhaul network allocates resources using the layered crossover optimization protocol, The wireless access network and the wired network network is a layer-optimized wireless mesh network implementation method characterized in that receiving the optimized service using the TCP / UDP protocol in consideration of the Internet connection and the user and the connection. The wireless backhaul network of claim 4, wherein: An ingress mesh router that analyzes the headers of packets and manages the queues, wherein the ingress mesh routers include queues in the elastic queue and the non-elastic queue. And separating TCP data and non-TCP data by analyzing packet headers, storing TCP data in the elastic queue, and storing the non-TCP data in the non-elastic queue. Optimized wireless mesh network implementation. The method of claim 5, wherein the ingress mesh router, The layer-optimized wireless mesh network, wherein the elastic queue is implemented as a per-destination queue or a per-link queue according to an implementation of a node-centric algorithm or a link-centric algorithm, to perform load sharing. Implementation method. The method of claim 5, wherein the wireless backhaul network, A link for controlling the non-elastic queue to be serviced before the elastic queue, wherein the non-elastic queue allocates a static bandwidth and the elastic queue determines which link to use A method of implementing a layer-optimized wireless mesh network, which allocates dynamic bandwidth according to an algorithm. The wireless backhaul network of claim 7, In order to match the existing source algorithm (TCP) and the hierarchical crossover optimization algorithm, TCP data is first stored in a TCP adaptation queue before being stored in the elastic queue. Way. The wireless backhaul network of claim 8, wherein: Hierarchically optimized by controlling the source algorithm of the hierarchical crossover optimization algorithm to be the same as the TCP rate by transmitting a simulated response according to the service rate during data transmission from the TCP adaptation queue to the elastic queue. How to implement a wireless mesh network.

Images