EP3580898A1

EP3580898A1 - Method and device for determining an energy-saving routing path

Info

Publication number: EP3580898A1
Application number: EP18701196.0A
Authority: EP
Inventors: Siwar BEN HADJ SAID; Alexandre Petrescu; Christophe JEANNETEAU
Original assignee: Commissariat a lEnergie Atomique CEA; Commissariat a lEnergie Atomique et aux Energies Alternatives CEA
Current assignee: Commissariat a lEnergie Atomique et aux Energies Alternatives CEA
Priority date: 2017-02-13
Filing date: 2018-01-30
Publication date: 2019-12-18
Also published as: FR3062976A1; FR3062976B1; WO2018145945A1

Abstract

The present invention relates to a method and a device for determining a data-routing path that can be used to reduce total energy consumption in networks. The methods takes account of both the available bandwidth of the path and the energy consumption of the links, without reducing the quality of service offered to users.

Description

METHOD AND DEVICE FOR DETERMINING ENERGY-EFFICIENT ROUTING PATH

Field of the invention

The invention lies in the field of computer and telecommunications networks and relates to optimizing the routing of data flows in an IP network.

State of the art

Recently, energy efficiency has become a key factor in the design of cellular networks, especially cellular networks ^5th generation (5G). This need is driven by a desire to reduce the carbon footprint of communications, minimize the grid operator's electricity bills, and extend the life of equipment running on batteries. In particular, this requirement is of paramount importance for network operators whose network infrastructure is constantly expanding by integrating new access technologies and increasing the number of access points to satisfy the growing number of clients. .

Several studies predict spectacular growth in Internet traffic due to streaming media services such as IPTV, Video on Demand (VoD) and video conferencing. Indeed, these services require routing paths with sufficient bandwidth that ensures low jitter and little loss of packets. Therefore, one of the biggest challenges in 5G networks is finding ways to reduce power consumption without compromising other performance metrics such as user throughput, jitter, or percentage of packets. lost. Considering all these parameters and constraints, routing algorithms become complex with multiple objectives and constraints. Energy efficiency in 5G access and aggregation networks is therefore a bi-objective optimization problem that must take into account both the available bandwidth and the energy consumption of the links, instead of just consider the bandwidth.

In addition, the European Commission has issued a Code of Conduct (CoC) on the energy consumption of broadband equipment that indicates the maximum power consumption allowed for each link technology. According to this report, each technology has its own energy consumption regardless of the associated bandwidth. For example, a point-to-point fiber with a bandwidth of 10 Gbps consumes about 8 Watts, while an EPON fiber with the same bandwidth consumes about 13.4 Watts.

Figures 1a and 1b illustrate for the same network configuration (100), the power that is consumed on a routing path according to which criteria are taken into account to determine the routing path. In this simplified example, the network (100) type 5G is composed of two parts: a so-called access network and aggregation portion (102), and a so-called core network portion (104). The two parts of the network are coupled via a router (103) also called edge node. The network comprises in each part, nodes or routers (106-1, 106-i, .... 106-n) which are connected to one another via links (108-1, 108-j, ... 108- m) may be of different technologies.

When transmitting or receiving packets, each router involved in a routing path develops a power consumption on its network interface that is a function of the technology of the link. The values indicated in FIGS. 1a and 1b are given by way of example, and corresponding to actual values given in the aforementioned code of conduct, for fiber types of 1 Gbit / s and 10 Gbit / s, depending on the number of ports. A client terminal (1 10) is connected to a WiFi terminal (1 12). A user requests a streamed video from a server (1 14) of the content provider. In the case of Figure 1a, the packet routing (path 1 1 6) is done based on the metric bandwidth. Thus, the links with the largest passerby are favored. This routing policy entails a total energy consumption of the path of 132W, corresponding to the sum of the energy consumption of each link involved in the path between the source and the destination. In the case of Figure 1b, the packet routing (path 1 18) is made according to the bandwidth and the energy consumption of the links. Thus, links that provide adequate bandwidth and consume less in terms of energy are favored. This routing policy entails a total energy consumption of the path of 74W.

This example illustrates the problem related to the need to have a routing protocol that reduces energy consumption.

There are two main approaches to meeting the challenge of having a routing protocol that can reduce energy consumption. The first approach is to propose a new routing metric that considers energy related parameters, such as the remaining level of a device's battery. A routing algorithm calculates the paths using this metric. The second approach is to introduce changes to the routing protocol itself in order to achieve energy savings. In this second approach, the routing metric does not undergo any modification.

With regard to the first approach, the existing solutions target networks of nodes with limited energy power such as ad-hoc networks or sensors. Indeed, in such a context, the node energy is a critical resource that must be used effectively to increase network life. Therefore, a routing metric based on node battery level or link quality is most appropriate for preserving node energy.

U.S. Patent Application 2013/0315257 A1 to Welin et al. presents a method that uses the power consumption metric when switching traffic in routers or switches. This metric can be based on the power consumption in the link interfaces, in the links and in the node. Each node receives energy consumption from the rest of the nodes that are in the same routing domain. It also sends its energy consumption to other nodes. This type of solution does not ensure a Quality of Service (QoS) required by the applications. Considering only energy as the only end-to-end selection criterion, this solution can choose to route latency-sensitive traffic (for example, emergency call, IPTV, etc.) by a path not respecting their need (for example, long path causing high latency). Although this approach makes it possible to reduce the energy consumption of the network, having considered the energy consumption of the interfaces as metrics, it does not make it possible to ensure the quality of service required by the applications, in particular the multimedia applications which have a very strong need. specific in terms of throughput, delays and percentages of lost packets.

U.S. Patent 6,301,244 B1 to Huang et al. describes a method for finding a path between a source node and each node in a communication network, so that the path delay does not exceed a path delay constraint or first metric and the cost of the path or second metric is minimized, each metric being additive. A first path that is the shortest path from a source node to each network node in terms of a first metric is selected using the Dijkstra algorithm. Nodes associated with paths whose metric exceeds the imposed constraint are eliminated. An accessibility graph is then constructed based on the remaining nodes. The Dijkstra algorithm is then applied to the accessibility graph to determine the shortest path between a source node and each node in terms of a second metric. The limitations of this approach are that it only targets the use of additive metrics, not concave metrics such as bandwidth. In addition, the fact of imposing a constraint on the first metric may, in the case where all the paths have latencies which exceed the maximum latency required, lead that the Dijkstra algorithm does not propose any first path, all the nodes being eliminated. . There is a risk of not finding a route to route a flow between a source and a destination.

With respect to the second approach, the basic idea is to route the data packets over a given subset of network links during periods of low traffic, using a coordination strategy between the routers. In this way, interfaces that do not participate in traffic switching can be turned off, in standby mode, without causing problems with network availability.

Zhang's US Patent Application 2015/0222537 A1 provides a method for aggregating traffic to disable or suspend network interfaces and thereby use as few links as possible in the network, thereby reducing the overall consumption of the network. network. This method uses a linear programming algorithm whose objective is to maximize the number of links to put on standby and taking as constraint the traffic state in the nodes of the network. Thus, for the smooth running of the algorithm, the method needs the following parameters: the source node, the destination node, data on the topology of the network as well as data on the state of the traffic in each node. The method considers that the energy consumption of interfaces is the same regardless of the technology used. In addition, the method is not compatible with Open Shortest Path First (OSPF) distributed routing protocols. Indeed, once the interface is disabled, the OSPF protocol considers that the link is down and will flood the network with messages "Link-state advertisements" (LSA) to inform other routers.

There is then the need for an appropriate solution to determine for each stream to be routed, a routing path that both meets the needs in terms of quality of service of the stream while minimizing the overall energy consumption of the network.

The present invention meets this need.

Summary of the invention

An object of the present invention is to propose a routing path determination method which reduces the total energy consumption in the networks, taking into account both the available bandwidth of the path and the energy consumption of the networks. nodes, and without degrading the quality of service offered to users. The present invention has the advantage of being able to be implemented on concave metrics, for example for the bandwidth where the metric of the end-to-end path is the minimum of the metric of the links, or to be implemented on metrics additives, for example for energy consumption where the end-to-end path metric is the sum of the link metrics.

Another object of the present invention is to propose a method of determining an energy-efficient routing path while preserving the quality of service constraints required by the data streams, whatever the heterogeneity of the technologies of the networks used. In general, the method of the invention relies on the implementation of an algorithm that comprises two levels of filtering: a first filtering to target all the best possible paths according to a first priority metric followed by a second filtering to select the best path according to a second metric.

The invention will find advantageous applications in all areas where telecommunication network providers and operators are eager to reduce the energy consumption of the networks while ensuring the quality of service required by the data flows to transit. Thus without limitation, the invention can be implemented:

- in the field of 5G cellular networks for any operator or equipment manufacturer;

- in the field of video distribution networks;

- in the field of real-time event networks (shows);

- in the field of home networks;

- in the field of wireless networks for industrial applications, including networks with battery-powered equipment;

- potentially for industrial providers of wireless communication solutions without infrastructure for public security or logistics.

Advantageously, the method of the invention is easy to implement, and adaptable according to the needs of the operator, depending on the priority performance sought, whether it is to ensure the quality of service for the flows, or to reduce energy consumption. .

If the priority is to ensure the quality of service for the streams, for example for a video stream, the method makes it possible to determine the paths having the best available bandwidth, and then to select among these, the one presenting the consumption. lowest energy level. If the priority is to reduce the energy consumption, in the case for example of flows that do not have a specific need in terms of quality of services as certain types of communications related to connected objects, the method makes it possible to determine the paths that have the lowest energy consumption, then select from those the one with the best available bandwidth.

In a preferred implementation, for calculating the best paths according to a given performance metric, the method implements a Yen algorithm, which makes it possible to find the best k paths without a loop.

To obtain the desired results, it is proposed a method of determining a data packet routing path between a source node and a destination node of a network comprising a plurality of nodes and links between the nodes, the links having at least first and second performance metrics, the method comprising the steps of:

calculating, according to the first performance metric on a representative graph of the nodes of the network, a set of candidates representing the best paths without a loop between the source node and the destination node, a path including at least one link, the set of candidates containing one or more candidates;

- classify the candidates according to the second performance metric; and

selecting a candidate according to the result of the ranking step, said selected candidate representing the routing path of the data packets between the source node and the destination node.

According to embodiments:

the ranking step consists in ordering the candidates according to an increasing or decreasing value of the second performance metric, and the selection step consists in selecting the candidate presenting the highest or lowest value; the calculation step consists in operating an algorithm for calculating 'k' shortest paths without a loop on a graph representative of the nodes of the network, 'k' being a predefined non-zero integer;

the algorithm is the Yen algorithm;

- the performance metrics are the available bandwidth and the energy consumption of each link;

the first metric is the available bandwidth and the second metric is the energy consumption;

the calculation step consists in determining a set of loopless paths having the highest available bandwidth and the selection step consists in selecting the path having the lowest energy consumption between the source node and the destination node;

- the first metric is the energy consumption and the second metric is the available bandwidth;

the calculation step consists in determining a set of loopless paths having the lowest power consumption and the selection step consists in selecting the path presenting the highest available bandwidth between the source node and the destination node;

the method further comprises, after the calculation step, a step for filtering among the candidates a subset of candidates according to quality of service parameters for the data stream;

the quality of service parameters are chosen from among the bit rate, the latency, the maximum loss rate of data packets, the jitter.

The invention covers a data packet routing path determining device between a source node and a destination node of a network comprising a plurality of nodes and links between the nodes, the links having at least first and second links. metrics of performance, the device comprising means for implementing the steps of the claimed method.

According to the embodiments:

- the links between the nodes of the network are wired and / or not wired; the data packets are packets of a multimedia data stream.

The invention also covers a communication system able to route in a network data packets between a source node and a destination node among a plurality of nodes connected by links, the links having at least first and second performance metrics, the system comprising at least one device as claimed.

In one embodiment, the claimed device is implemented in at least one node of the network.

In another embodiment, the network is of the "Software-

Defined Networking "(SDN) including an SDN controller, and the claimed device is implemented in the SDN controller of that network.

The invention also covers a method operated by an SDN controller in a Software-Challenge Networking (SDN) network, for constructing a data flow route, the network comprising a plurality of nodes and links between the nodes. the links having at least first and second performance metrics, the SDN controller being adapted to detect a stream routing request and identify the address of a source node and the address of a destination node for the stream, the method comprising the steps of the routing path determination method as claimed.

According to embodiments, the execution of the steps of the data packet routing path determination method is done according to whether the quality of service for a data stream has priority or not The invention may operate in the form of a computer program product that includes code instructions for performing the claimed process steps when the program is run on a computer.

Description of figures

Various aspects and advantages of the invention will appear in support of the description of a preferred embodiment of the invention, but not limiting, with reference to the figures below:

Figures 1a and 1b illustrate the power that is consumed on a routing path according to which criteria are taken into account to determine the routing path;

FIG. 2 illustrates a sequence of steps according to the method of the invention, to determine a routing path;

Figure 3 details the steps of the method of the invention in one embodiment;

Figure 4 details the steps of the method of the invention in another embodiment;

Figure 5 illustrates a sequence of steps according to a variant of the method of the invention in another embodiment; Figure 6 illustrates a sequence of steps of the method of the invention in one embodiment for an SDN network;

Figure 7 schematically shows an SDN environment in which the invention can be implemented.

DETAILED DESCRIPTION OF THE INVENTION A major advantage of the process of the invention, also referred to as the "GoGreen" process, is to reduce energy consumption in networks when calculating a routing path, taking into account both the available bandwidth of links between nodes and their energy consumption.

Figure 2 illustrates a sequence of steps (200) of the GoGreen process of the invention. The method makes it possible to determine a routing path in communication networks, such as, for example, the 5G cellular network as illustrated in FIGS. 1a and 1b. The Gogreen method can also be operated in networks implementing a distributed routing protocol, such as the OSPF protocol. In the latter case, the method is executed in each router of the network, and the router block "Router Link State Advertisment" (Router LSA) in the message "Link State Update" (LSU) which is used for each router announces its interfaces , its neighbors and metric values to reach these neighbors, is modified to include a field announcing the two metrics of bandwidth and energy consumption.

To initiate the process, several parameters are required, namely:

a graphical representation of the nodes of the network where each link between the nodes is characterized by two performance metrics (w1, w2);

a source-destination pair (src, dst) corresponding to a transmitter of a data stream and a receiver of the stream; and

a maximum number 'k' of the best paths to be computed, the parameter 'k' being a predefined non-zero integer.

Typically, 'k' varies between 1 and 5 for a graph with nodes that vary between 10 and 100.

In a first step (202), the method makes it possible to determine a non-zero set of candidates of 'k' paths between the source and the destination presenting the first performance metric 'w1', without impose a minimum constraint or restriction on this first metric.

In the case where only one candidate is proposed, the method selects (206) directly the corresponding path.

In the case where several candidates are obtained from the above calculation, the method makes it possible to classify (204) the candidates according to the second performance metric 'w2'. The ranking step consists of ordering the 'k' candidates according to an increasing or decreasing value of the second performance metric.

After the classification step, the method makes it possible to select

(206) a candidate with the best value for the second metric w2. The selection step consists of selecting the candidate with the highest or lowest value in the ranking to enable the transmission of the data stream between the source and the destination via this path. The selected candidate then corresponds to the path offering optimized performance with respect to the two metrics.

In an implementation, the step of determining paths I having the best values of the performance metric considered further comprises the classification step, a step of sorting the candidates to keep as candidates only the paths without loop.

In a preferred implementation, the step of determining the best paths according to a performance metric is based on the well-known Yen algorithm (JK Yen, "Finding the Shortest Loop Paths in a Network", Management Science (Vol 17 , No. 1 1, pp. 712-71 6), 1971) which makes it possible to generate the 'k' shortest paths without loops in a graph, and the classification step is done directly on the candidates without loop. FIG. 3 details the steps (300) of the method of the invention, in one embodiment (302) where the first metric 'w1 = BP' is the available bandwidth, and the second metric 'w2 = PW is the consumption energy. Such an implementation corresponds to the scenarios where the performance priority is given to the bandwidth.

The GoGreen method calculates (step 304) a non-zero set of k best paths by considering as first metric the available bandwidth (w1) of the links, without imposing a minimum constraint or a restriction on this first metric. It determines the k loopless paths that have the highest available bandwidth, the bandwidth of a path being considered to be equal to the minimum bandwidths of the links that make up that path. If only one path is candidate (OUI branch of 306), the method makes it possible to select it (31 0) to make it possible to activate the transmission of the data flow between the source and the destination via this path. If several paths are candidates (branch NO of 306), the method makes it possible to classify (308) the candidates according to the value of the consumption

end-to-end energy consumption, the energy consumption of a path being calculated by summing the energy consumption of the links composing the path. In the next step (31 0), the method makes it possible to select the path which has the smallest end-to-end energy consumption to enable the transmission of the data stream between the source and the destination via this process. path.

FIG. 4 details the steps (400) of the method of the invention, in one embodiment (402) where the first metric 'w1 = PW is the energy consumption, and the second metric' w2 = BP 'is available bandwidth . Such an implementation corresponds to the scenarios where the priority of performance is given to energy consumption. The GoGreen method calculates (step 404) a non-zero set of k best paths by considering as first metric the energy consumption (w1) of the links, without imposing a minimum constraint or a restriction on this first metric. It determines the k loopless paths that have the lowest end-to-end energy consumption, the energy consumption of a path being calculated by summing the energy consumption of the links making up the path. If only one path is candidate (YES branch of 406), the method allows to select it (410) to enable the transmission of data flow between the source and the destination via this path. If several paths are candidates (NOT branch of 406), the method makes it possible to classify (408) the candidates according to the value of the available bandwidth of each candidate, the bandwidth of a path being considered as being equal to the minimum of the bands passers of the links that make up this path. In the next step (410), the method makes it possible to select the path that has the largest bandwidth available to enable the transmission of the data stream between the source and the destination via this path.

In a preferred implementation, the calculation step (304, 404) applies a Yen-type algorithm that makes it possible to determine the k shortest paths without a loop.

Figure 5 illustrates a sequence of steps (500) of a variant of the method of the invention in an embodiment where the traffic requirements to be routed in terms of quality of service (QoS) are considered.

This variant is particularly applicable for the case where energy consumption is the priority performance. Indeed, the previously described GoGreen method operates without having to know beforehand the specific traffic requirements to be routed in terms of QoS (for example, the threshold rate required by the traffic of data). In the implementation according to FIG. 3, the method makes it possible to generate the k paths, where 'k' is a non-zero integer having the best available bandwidth, without imposing a minimum constraint or a restriction on this first metric. Then, it selects, among these k paths, the one with the best value of energy consumption. However, in this mode, it may happen that the path ultimately selected is a path that reduces energy consumption but which however does not ensure the quality of service required by the traffic to route. Indeed, when the method considers the specific requirements of each data stream during the determination of the candidates, it can best meet the different requirements of the traffic. For example, the method must determine a routing path for a video stream that requires a bit rate of 0.5 Mbps. Assuming 'k = 4', the process generates 4 paths (ch) with the best rates: ch1 = (5Mbits / s, 10Watts), ch2 = (2Mbits / s, 6Watts), ch3 = (1 Mbps) , 4Watts) and ch4 = (0.2Mbits / s, 2Watts). In the selection step, the method will select the fourth path ch4 to route the traffic since it has the lowest power consumption. However, the bit rate that the ch4 path can provide does not satisfy the specific traffic requirements in terms of bit rate (0.5Mbps).

An advantageous variant of the method of the invention in the embodiment according to FIG. 3 consists in taking into account the traffic requirements in the selection of the routing path. In this version, which considers as first metric the bandwidth of the links (w1) and as the second metric the energy consumption (w2), the method has a prior knowledge of the bit rate required by the data traffic to be routed (502). The method determines a group 'R' of k paths that offer the best bandwidth (504). If the group R contains only one path (YES branch of 506), the method selects (51 6) this path to initiate the transmission of data between the source and the destination. If several paths are candidates (branch NO of 506), the method makes it possible to select (508) from among the candidates, a subgroup 'R_QoS' of candidates that make it possible to ensure the traffic requirements. The traffic requirements may include selected quality of service parameters among rate, latency, maximum data packet loss rate, jitter.

If the candidate list of the R_QoS subgroup is empty (YES branch of 510), the method makes it possible to classify (512) the candidates of the initial group R according to the second metric of energy consumption, and to select (51 6) from the group R the path presenting the best value according to the second metric of energy consumption.

If the list of candidates of the subgroup R_QoS is not empty (branch NOT of 510), the method makes it possible to classify (514) the candidates of the subgroup R_QoS according to the second metric of energy consumption, and to select (516) among the group R_QoS the path presenting the best value according to the second metric of energy consumption.

In a particular embodiment of the invention, the QoS requirements of a data stream may be indicated in the headers of the packets of the stream. FIG. 6 illustrates a sequence of stream route construction steps (600) incorporating the routing path determination method of the invention, in one embodiment where the network is an SDN network, controlled by an SDN controller. . Fig. 7 schematically shows an SDN environment in which the method of Fig. 6 can be implemented at the SDN controller (702). One of the features of the SDN controller is its ability to have a global view of the network topology. This allows it to execute the method of the invention to calculate the optimal routing path while taking into account different objectives and constraints. The network (SDN) includes SDN switches for the transfer of traffic from the user in the data plane. These devices are controlled by the SDN controller. Advantageously, the implementation of the method of the invention at the level of the SDN controller does not imply any modification or adaptation in the SDN switches or in the exchange protocol (ie Openflow) between the controller and the switch. The network devices in the data plane (for example, routers, switches, radio cells, etc.) just need to be equipped with a protocol allowing the configuration of their routing table by the SDN controller. , as for example the OpenFlow protocol, while the calculation of the routing path is performed in the SDN controller, for example in a "data center" or a server having high performance.

The method (600) is triggered by the arrival of a new OpenFlow message "OF PACKETJN" from a given switch or router. This message allows the SDN controller to detect (602) the need to route a new stream within the network. From this message, the SDN controller identifies the source address, the destination address, the transport protocol used, the source and destination port number. In a next step, the method allows the SDN controller to determine (604) whether the QoS quality of service has priority based on various parameters such as the transport protocol used, the port number that identifies the application, and the destination address. If the QoS has priority (YES branch of 604), the method executes the routing path selection steps according to the operating mode described when the priority is the bandwidth (according to FIG. 3 or its variant according to FIG. 5). In the opposite case where the QoS is not a priority (NO branch of 604), the method executes the routing path selection steps according to the operating mode described when the priority is the energy consumption (according to FIG. 4). In each mode of operation, the SDN may decide to use different values of the parameter k for different flow subgroups.

The choice of the parameter k, as a non-zero integer, makes it possible to limit the number of the best paths selected according to the first metric, for reasons of performance. Indeed, the fact of calculating a precise number of possible paths (in this case k) is more efficient in terms of computing time than calculating all the possible paths. In the SDN context, the computation of the paths must be done in real time as soon as a new flow appears in the network, at the level of a router or input switch. This new stream is then detected by the SDN controller which must therefore quickly calculate the path for this stream and set up the corresponding routing configurations in the routers of the calculated path, in order to allow the routing of the stream. If the controller took too long to calculate a path with solutions that seek to calculate all the possible paths, such as the approach of the aforementioned US Patent 6,301, 244 B1 which furthermore comprises two successive applications of the calculation algorithm Dijkstra , there would then be a delay too long for setting up the configuration of the path in the network. This would result in a significant degradation of the quality of services for this flow with a long delay in the establishment of the flow and the end-to-end connection, a high risk that the flow is interrupted, a high rate of losses of packages, great latencies, among others.

Thus, the execution of the GoGreen process provides the routing path that satisfies the needs of the flow. The SDN controller determines (610) the switches / routers that are part of this routing path. Then, it prepares (612) the "OF PACKET_OUT" messages which make it possible to configure (614) these switches / routers.

Once the path is established in the data plane, the controller

SDN updates (61 6) the value of available bandwidth of links on which the stream was routed. When the transmission of the data flow is complete, the switch / router sends a message to the SDN controller to indicate the end of the transmission of the stream at its level. This allows the controller to update the bandwidth value available to the links. Indeed, in the OpenFlow protocol, each switch rule in a switch / router is associated with a timer that expires when the switch / router no longer receives packet linked to the stream. In this case, the switch rule is removed and a notification message is sent to the SDN controller.

Advantageously, the described stream route construction method makes it possible to coexist in the same network flows having different needs, by the execution of the method of the invention:

- in a mode where the priority is to ensure QoS, while seeking to reduce energy consumption; and

- in mode where there are no imposed QoS constraints but where it is important to minimize energy consumption.

It should be noted that the available bandwidth at initialization corresponds to the nominal bandwidth. After installing a new path for a given stream, the available bandwidth is then the nominal bandwidth minus the estimated load for that stream. In the case of the SDN, the SDN controller initializes the available bandwidth to the nominal bandwidth value. After installing a path in the data transfer plan for a given stream, the controller updates this metric by subtracting the estimated load for this stream, and this for each link participating in the routing of this stream. Thus, in the example of an available bandwidth of 10 Mbps and a path X composed of three links (11, 12, 13) respectively having bandwidths of (11 Mbps, 10 Mbps, 11 Mbps), after installing this path for a video stream that needs 2Mbps, the controller updates the bandwidth metric for each link in that path. So the bandwidth of the first link 11 becomes 9Mbps, that of the second link 12 becomes 8Mbps and that of the third link 13 becomes 9Mbps. The X path then has an available bandwidth of 9Mbps. To determine the available bandwidth of a path, it is sufficient to determine the minimum of the values of the available bandwidth of the links that constitute this path.

Those skilled in the art will consider that the present invention can be implemented from hardware and / or software elements and operate on a computer. It may be available as a computer program product on a computer readable medium. The support can be electronic, magnetic, optical, electromagnetic or be an infrared type of diffusion medium. Such supports are, for example, Random Access Memory RAMs (ROMs), magnetic or optical tapes, disks or disks (Compact Disk - Read Only Memory (CD-ROM), Compact Disk - Read / Write (CD-R / W) and DVD). Thus, the present description illustrates a preferred implementation of the invention, but is not limiting. An example has been chosen to allow a good understanding of the principles of the invention, and a concrete application, but it is in no way exhaustive and should allow the skilled person to make changes and implementation variants in keeping the same principles.

Claims

claims

A method of determining a data packet routing path between a source node and a destination node of a network comprising a plurality of nodes and links between the nodes, the links having at least first and second performance metrics, the method comprising the steps of:

calculating, according to the first performance metric on a representative graph of the nodes of the network, a set of candidates representing the best paths without a loop between the source node and the destination node, a path including at least one link, a set containing one or more candidates;

- classify the candidates according to the second performance metric; and

The method of claim 1 wherein the ranking step includes ordering the candidates to an increasing or decreasing value of the second performance metric, and the selecting step includes selecting the candidate with the highest or the highest value. low.

3. Method according to any one of claims 1 to 2 wherein the calculation step consists in operating a calculation algorithm of 'k' shortest paths without a loop on a representative graph of the nodes of the network, 'k' being a predefined integer.

The method of claim 3 wherein the algorithm is the Yen algorithm.

5. Method according to any one of claims 1 to 4 wherein the performance metrics are the available bandwidth and energy consumption of each link.

6. Method according to any one of claims 1 to 5 wherein the first metric is the available bandwidth and the second metric is the energy consumption. The method of claim 6 wherein the computing step is to determine a set of loopless paths having the highest available bandwidth and the selecting step is to select the path with the lowest power consumption between the source node and the destination node.

8. Method according to any one of claims 1 to 5 wherein the first metric is the energy consumption and the second metric is the available bandwidth.

9. The method of claim 8 wherein the computing step is to determine a set of paths with the lowest energy consumption and the selection step consists of select the path with the highest available bandwidth between the source node and the destination node.

The method of claim 6 further comprising after the computing step, a filtering step for filtering among candidates a subset of candidates according to quality of service parameters for the data stream.

1 1. The method of claim 10 wherein the quality of service parameters are selected from the throughput, latency, maximum data packet loss rate, jitter.

12. A computer program product, said computer program comprising code instructions for performing the steps of the method according to any one of claims 1 to 11, when said program is executed on a computer.

13. A data packet routing path determining device between a source node and a destination node of a network comprising a plurality of nodes and links between the nodes, the links having at least first and second performance metrics, the device comprising means for implementing the steps of the method according to any one of claims 1 to 1 1.

14. The device of claim 13 wherein the links between the nodes of the network are wired and / or non-wired.

The device of claim 13 or 14 wherein the data packets are packets of a multimedia data stream.

A communication system capable of routing in a network data packets between a source node and a destination node among a plurality of nodes connected by links, the links having at least first and second performance metrics, the system comprising at least one device according to any one of claims 13 to 15.

The system of claim 1 6 wherein said at least one device is implemented in at least one node of the network.

18. The system of claim 1 6 or 17 wherein the network is of the "Software-Defined Networking" (SDN) type and said at least one device is implemented in the SDN controller of said network.

19. Method of constructing a route by data flow in a network of the "Software-Defined Networking" (SDN) type, the network comprising a plurality of nodes and links between the nodes, the links having at least first and second metrics the network comprising an SDN controller adapted to detect a stream routing request and identify the address of a source node and the address of a destination node for the stream, the method being operated by said SDN controller and comprising the steps of the data packet routing path determining method between the source node and the destination node according to any one of claims 1 to 1 1.

The method of claim 19 wherein the execution of the steps of the data packet routing path determining method is performed according to whether or not the quality of service for a data stream is a priority.

21. A computer program product, said computer program comprising code instructions for performing the steps of the method of any of claims 19 to 20, when said program is run on a computer.