US20190363951A1

Movatterモバイル変換

Info

Publication number: US20190363951A1
Application number: US16/532,978
Authority: US
Inventors: Jean-Philippe Vasseur; Grégory Mermoud; Sukrit Dasgupta
Original assignee: Cisco Technology Inc
Current assignee: Cisco Technology Inc
Priority date: 2014-01-06
Filing date: 2019-08-06
Publication date: 2019-11-28
Also published as: US20150195144A1; US9369351B2; US20150195136A1; US20150195212A1; US10277476B2; US20150195126A1; US20150195184A1; US9485153B2; US10425294B2; US9426040B2; US9374281B2; US9473364B2; US20150195216A1; US20150195185A1

Abstract

In one embodiment, one or more reporting nodes are selected to report network metrics in a network. From a monitoring node in the network, a trigger message is sent to the one or more reporting nodes. The trigger message may trigger the one or more reporting nodes to report one or more network metrics local to the respective reporting node. In response to the trigger message, a report of the one or more network metrics is received at the monitoring node from one of the one or more reporting nodes.

Description

RELATED APPLICATION

This application is a continuation of U.S. patent application Ser. No. 14/164,444, filed Jan. 27, 2014, entitled DISTRIBUTED AND LEARNING MACHINE-BASED APPROACH TO GATHERING LOCALIZED NETWORK DYNAMICS, which claims priority to U.S. Provisional Application No. 61/923,910, filed Jan. 6, 2014, entitled: LEARNING-MACHINE-BASED PREDICTIVE AND PROACTIVE COMPUTER NETWORKING AND ASSOCIATED MONITORING, both by Vasseur, et al., the contents of each of which are herein incorporated by reference.

TECHNICAL FIELD

The present disclosure relates generally to computer networks, and, more particularly, to the use of learning machines within computer networks.

BACKGROUND

Low power and Lossy Networks (LLNs), e.g., Internet of Things (IoT) networks, have a myriad of applications, such as sensor networks, Smart Grids, and Smart Cities. Various challenges are presented with LLNs, such as lossy links, low bandwidth, low quality transceivers, battery operation, low memory and/or processing capability, etc. The challenging nature of these networks is exacerbated by the large number of nodes (an order of magnitude larger than a “classic” IP network), thus making the routing, Quality of Service (QoS), security, network management, and traffic engineering extremely challenging, to mention a few.

Machine learning (ML) is concerned with the design and the development of algorithms that take as input empirical data (such as network statistics and performance indicators), and recognize complex patterns in these data. In general, these patterns are then used to make decisions automatically (i.e., close-loop control) or to help make decisions. ML is a very broad discipline used to tackle very different problems (e.g., computer vision, robotics, data mining, search engines, etc.), but the most common tasks are the following: linear and non-linear regression, classification, clustering, dimensionality reduction, anomaly detection, optimization, association rule learning.

One very common pattern among ML algorithms is the use of an underlying model M, whose parameters are optimized for minimizing the cost function associated to M, given the input data. For instance, in the context of classification, the model M may be a straight line that separates the data into two classes such that M =a*x+b*y+c and the cost function would be the number of misclassified points. The ML algorithm then consists in adjusting the parameters a,b,c such that the number of misclassified points is minimal. After this optimization phase (or learning phase), the model M can be used very easily to classify new data points. Often, M is a statistical model, and the cost function is inversely proportional to the likelihood of M, given the input data.

Learning Machines (LMs) are computational entities that rely one or more ML algorithm for performing a task for which they haven't been explicitly programmed to perform. In particular, LMs are capable of adjusting their behavior to their environment. In the context of LLNs, and more generally in the context of the IoT (or Internet of Everything, IoE), this ability will be very important, as the network will face changing conditions and requirements, and the network will become too large for efficiently management by a network operator.

Thus far, LMs have not generally been used in LLNs, despite the overall level of complexity of LLNs, where “classic” approaches (based on known algorithms) are inefficient or when the amount of data cannot be processed by a human to predict network behavior considering the number of parameters to be taken into account.

BRIEF DESCRIPTION OF THE DRAWINGS

The embodiments herein may be better understood by referring to the following description in conjunction with the accompanying drawings in which like reference numerals indicate identically or functionally similar elements, of which:

FIG. 1 illustrates an example communication network;

FIG. 2 illustrates an example network device/node;

FIG. 3 illustrates an example directed acyclic graph (DAG) in the communication network ofFIG. 1;

FIG. 4 illustrates an example Bayesian network;

FIG. 5 illustrates an example Bayesian network for linear regression; and

FIG. 6 illustrates an example simplified procedure for a distributed and learning machine-based approach to gathering localized network dynamics.

DESCRIPTION OF EXAMPLE EMBODIMENTSOverview

According to one or more embodiments of the disclosure, one or more reporting nodes are selected by a Learning Machine to report network metrics in a network. From a monitoring node in the network, a trigger message is sent to the one or more reporting nodes. The trigger message may trigger the one or more reporting nodes to report one or more network metrics local to the respective reporting node, which may be used to report ephemeral phenomena. In response to the trigger message, a report of the one or more network metrics is received at the monitoring node from one of the one or more reporting nodes.

Description

A computer network is a geographically distributed collection of nodes interconnected by communication links and segments for transporting data between end nodes, such as personal computers and workstations, or other devices, such as sensors, etc. Many types of networks are available, ranging from local area networks (LANs) to wide area networks (WANs). LANs typically connect the nodes over dedicated private communications links located in the same general physical location, such as a building or campus. WANs, on the other hand, typically connect geographically dispersed nodes over long-distance communications links, such as common carrier telephone lines, optical lightpaths, synchronous optical networks (SONET), synchronous digital hierarchy (SDH) links, or Powerline Communications (PLC) such as IEEE 61334, IEEE P1901.2, and others. In addition, a Mobile Ad-Hoc Network (MANET) is a kind of wireless ad-hoc network, which is generally considered a self-configuring network of mobile routers (and associated hosts) connected by wireless links, the union of which forms an arbitrary topology.

Smart object networks, such as sensor networks, in particular, are a specific type of network having spatially distributed autonomous devices such as sensors, actuators, etc., that cooperatively monitor physical or environmental conditions at different locations, such as, e.g., energy/power consumption, resource consumption (e.g., water/gas/etc. for advanced metering infrastructure or “AMI” applications) temperature, pressure, vibration, sound, radiation, motion, pollutants, etc. Other types of smart objects include actuators, e.g., responsible for turning on/off an engine or perform any other actions. Sensor networks, a type of smart object network, are typically shared-media networks, such as wireless or PLC networks. That is, in addition to one or more sensors, each sensor device (node) in a sensor network may generally be equipped with a radio transceiver or other communication port such as PLC, a microcontroller, and an energy source, such as a battery. Often, smart object networks are considered field area networks (FANs), neighborhood area networks (NANs), personal area networks (PANs), etc. Generally, size and cost constraints on smart object nodes (e.g., sensors) result in corresponding constraints on resources such as energy, memory, computational speed and bandwidth.

FIG. 1 is a schematic block diagram of anexample computer network100 illustratively comprising nodes/devices110 (e.g., labeled as shown, “root,” “11,” “12,” . . . “45,” and described inFIG. 2 below) interconnected by various methods of communication. For instance, thelinks105 may be wired links or shared media (e.g., wireless links, PLC links, etc.) wherecertain nodes110, such as, e.g., routers, sensors, computers, etc., may be in communication withother nodes110, e.g., based on distance, signal strength, current operational status, location, etc. The illustrative root node, such as a field area router (FAR) of a FAN, may interconnect the local network with aWAN130, which may house one or more other relevant devices such as management devices orservers150, e.g., a network management server (NMS), a dynamic host configuration protocol (DHCP) server, a constrained application protocol (CoAP) server, etc. Those skilled in the art will understand that any number of nodes, devices, links, etc. may be used in the computer network, and that the view shown herein is for simplicity. Also, those skilled in the art will further understand that while the network is shown in a certain orientation, particularly with a “root” node, thenetwork100 is merely an example illustration that is not meant to limit the disclosure.

Data packets140 (e.g., traffic and/or messages) may be exchanged among the nodes/devices of thecomputer network100 using predefined network communication protocols such as certain known wired protocols, wireless protocols (e.g., IEEE Std. 802.15.4, WiFi, Bluetooth®, etc.), PLC protocols, or other shared-media protocols where appropriate. In this context, a protocol consists of a set of rules defining how the nodes interact with each other.

FIG. 2 is a schematic block diagram of an example node/device200 that may be used with one or more embodiments described herein, e.g., as any of the nodes or devices shown inFIG. 1 above. The device may comprise one or more network interfaces210 (e.g., wired, wireless, PLC, etc.), at least oneprocessor220, and amemory240 interconnected by a system bus250, as well as a power supply260 (e.g., battery, plug-in, etc.).

The network interface(s)210 contain the mechanical, electrical, and signaling circuitry for communicating data overlinks105 coupled to thenetwork100. The network interfaces may be configured to transmit and/or receive data using a variety of different communication protocols. Note, further, that the nodes may have two different types ofnetwork connections210, e.g., wireless and wired/physical connections, and that the view herein is merely for illustration. Also, while thenetwork interface210 is shown separately frompower supply260, for PLC (where the PLC signal may be coupled to the power line feeding into the power supply) thenetwork interface210 may communicate through thepower supply260, or may be an integral component of the power supply.

Thememory240 comprises a plurality of storage locations that are addressable by theprocessor220 and thenetwork interfaces210 for storing software programs and data structures associated with the embodiments described herein. Note that certain devices may have limited memory or no memory (e.g., no memory for storage other than for programs/processes operating on the device and associated caches). Theprocessor220 may comprise hardware elements or hardware logic adapted to execute the software programs and manipulate thedata structures245. Anoperating system242, portions of which are typically resident inmemory240 and executed by the processor, functionally organizes the device by, inter alia, invoking operations in support of software processes and/or services executing on the device. These software processes and/or services may comprise a routing process/services244 and an illustrative “learning machine”process248, which may be configured depending upon the particular node/device within thenetwork100 with functionality ranging from intelligent learning machine algorithms to merely communicating with intelligent learning machines, as described herein. Note also that while thelearning machine process248 is shown incentralized memory240, alternative embodiments provide for the process to be specifically operated within thenetwork interfaces210.

It will be apparent to those skilled in the art that other processor and memory types, including various computer-readable media, may be used to store and execute program instructions pertaining to the techniques described herein. Also, while the description illustrates various processes, it is expressly contemplated that various processes may be embodied as modules configured to operate in accordance with the techniques herein (e.g., according to the functionality of a similar process). Further, while the processes have been shown separately, those skilled in the art will appreciate that processes may be routines or modules within other processes.

Routing process (services)244 contains computer executable instructions executed by theprocessor220 to perform functions provided by one or more routing protocols, such as proactive or reactive routing protocols as will be understood by those skilled in the art. These functions may, on capable devices, be configured to manage a routing/forwarding table (a data structure245) containing, e.g., data used to make routing/forwarding decisions. In particular, in proactive routing, connectivity is discovered and known prior to computing routes to any destination in the network, e.g., link state routing such as Open Shortest Path First (OSPF), or Intermediate-System-to-Intermediate-System (ISIS), or Optimized Link State Routing (OLSR). Reactive routing, on the other hand, discovers neighbors (i.e., does not have an a priori knowledge of network topology), and in response to a needed route to a destination, sends a route request into the network to determine which neighboring node may be used to reach the desired destination. Example reactive routing protocols may comprise Ad-hoc On-demand Distance Vector (AODV), Dynamic Source Routing (DSR), DYnamic MANET On-demand Routing (DYMO), etc. Notably, on devices not capable or configured to store routing entries, routingprocess244 may consist solely of providing mechanisms necessary for source routing techniques. That is, for source routing, other devices in the network can tell the less capable devices exactly where to send the packets, and the less capable devices simply forward the packets as directed.

Notably, mesh networks have become increasingly popular and practical in recent years. In particular, shared-media mesh networks, such as wireless or PLC networks, etc., are often on what is referred to as Low-Power and Lossy Networks (LLNs), which are a class of network in which both the routers and their interconnect are constrained: LLN routers typically operate with constraints, e.g., processing power, memory, and/or energy (battery), and their interconnects are characterized by, illustratively, high loss rates, low data rates, and/or instability. LLNs are comprised of anything from a few dozen and up to thousands or even millions of LLN routers, and support point-to-point traffic (between devices inside the LLN), point-to-multipoint traffic (from a central control point such at the root node to a subset of devices inside the LLN) and multipoint-to-point traffic (from devices inside the LLN towards a central control point).

An example implementation of LLNs is an “Internet of Things” network. Loosely, the term “Internet of Things” or “IoT” (or “Internet of Everything” or “IoE”) may be used by those in the art to refer to uniquely identifiable objects (things) and their virtual representations in a network-based architecture. In particular, the next frontier in the evolution of the Internet is the ability to connect more than just computers and communications devices, but rather the ability to connect “objects” in general, such as lights, appliances, vehicles, HVAC (heating, ventilating, and air-conditioning), windows and window shades and blinds, doors, locks, etc. The “Internet of Things” thus generally refers to the interconnection of objects (e.g., smart objects), such as sensors and actuators, over a computer network (e.g., IP), which may be the Public Internet or a private network. Such devices have been used in the industry for decades, usually in the form of non-IP or proprietary protocols that are connected to IP networks by way of protocol translation gateways. With the emergence of a myriad of applications, such as the smart grid, smart cities, and building and industrial automation, and cars (e.g., that can interconnect millions of objects for sensing things like power quality, tire pressure, and temperature and that can actuate engines and lights), it has been of the utmost importance to extend the IP protocol suite for these networks.

An example protocol specified in an Internet Engineering Task Force (IETF) Proposed Standard, Request for Comment (RFC) 6550, entitled “RPL: IPv6 Routing Protocol for Low Power and Lossy Networks” by Winter, et al. (March 2012), provides a mechanism that supports multipoint-to-point (MP2P) traffic from devices inside the LLN towards a central control point (e.g., LLN Border Routers (LBRs), FARs, or “root nodes/devices” generally), as well as point-to-multipoint (P2MP) traffic from the central control point to the devices inside the LLN (and also point-to-point, or “P2P” traffic). RPL (pronounced “ripple”) may generally be described as a distance vector routing protocol that builds a Directed Acyclic Graph (DAG) for use in routing traffic/packets140, in addition to defining a set of features to bound the control traffic, support repair, etc. Notably, as may be appreciated by those skilled in the art, RPL also supports the concept of Multi-Topology-Routing (MTR), whereby multiple DAGs can be built to carry traffic according to individual requirements.

Also, a directed acyclic graph (DAG) is a directed graph having the property that all edges are oriented in such a way that no cycles (loops) are supposed to exist. All edges are contained in paths oriented toward and terminating at one or more root nodes (e.g., “clusterheads or “sinks”), often to interconnect the devices of the DAG with a larger infrastructure, such as the Internet, a wide area network, or other domain. In addition, a Destination Oriented DAG (DODAG) is a DAG rooted at a single destination, i.e., at a single DAG root with no outgoing edges. A “parent” of a particular node within a DAG is an immediate successor of the particular node on a path towards the DAG root, such that the parent has a lower “rank” than the particular node itself, where the rank of a node identifies the node's position with respect to a DAG root (e.g., the farther away a node is from a root, the higher is the rank of that node). Note also that a tree is a kind of DAG, where each device/node in the DAG generally has one parent or one preferred parent. DAGs may generally be built (e.g., by a DAG process and/or routing process244) based on an Objective Function (OF). The role of the Objective Function is generally to specify rules on how to build the DAG (e.g. number of parents, backup parents, etc.).

FIG. 3 illustrates an example simplified DAG that may be created, e.g., through the techniques described above, withinnetwork100 ofFIG. 1. For instance,certain links105 may be selected for each node to communicate with a particular parent (and thus, in the reverse, to communicate with a child, if one exists). These selected links form the DAG310 (shown as bolded lines), which extends from the root node toward one or more leaf nodes (nodes without children). Traffic/packets140 (shown inFIG. 1) may then traverse theDAG310 in either the upward direction toward the root or downward toward the leaf nodes, particularly as described herein.

RPL supports two modes of operation for maintaining and using Downward routes:

1) Storing Mode: RPL routers unicast DAO messages directly to their DAG Parents. In turn, RPL routers maintain reachable IPv6 addresses for each of their DAG Children in their routing table. Because intermediate RPL routers store Downward routing state, this mode is called Storing mode.

2) Non-Storing Mode: RPL routers unicast DAO messages directly to the DAG Root. The DAO message also includes the IPv6 addresses for the source's DAG Parents. By receiving DAO messages from each RPL router in the network, the DAG Root obtains information about the DAG topology and can use source routing to deliver datagrams. Unlike Storing mode, intermediate RPL routers in Non-Storing mode do not maintain any Downward routes.

Learning Machine Technique(s)

As noted above, machine learning (ML) is concerned with the design and the development of algorithms that take as input empirical data (such as network statistics and performance indicators), and recognize complex patterns in these data. One very common pattern among ML algorithms is the use of an underlying model M, whose parameters are optimized for minimizing the cost function associated to M, given the input data. For instance, in the context of classification, the model M may be a straight line that separates the data into two classes such that M =a*x +b*y +c and the cost function would be the number of misclassified points. The ML algorithm then consists in adjusting the parameters a,b,c such that the number of misclassified points is minimal. After this optimization phase (or learning phase), the model M can be used very easily to classify new data points. Often, M is a statistical model, and the cost function is inversely proportional to the likelihood of M, given the input data.

As also noted above, learning machines (LMs) are computational entities that rely one or more ML algorithm for performing a task for which they haven't been explicitly programmed to perform. In particular, LMs are capable of adjusting their behavior to their environment. In the context of LLNs, and more generally in the context of the IoT (or Internet of Everything, IoE), this ability will be very important, as the network will face changing conditions and requirements, and the network will become too large for efficiently management by a network operator. Thus far, LMs have not generally been used in LLNs, despite the overall level of complexity of LLNs, where “classic” approaches (based on known algorithms) are inefficient or when the amount of data cannot be processed by a human to predict network behavior considering the number of parameters to be taken into account.

In particular, many LMs can be expressed in the form of a probabilistic graphical model also called Bayesian Network (BN). A BN is a graph G=(V,E) where V is the set of vertices and E is the set of edges. The vertices are random variables, e.g., X, Y, and Z (seeFIG. 4) whose joint distribution P(X,Y,Z) is given by a product of conditional probabilities:

P(X,Y,Z)=P(Z|X,Y)P(Y|X)P(X) (Eq. 1)

The conditional probabilities in Eq. 1 are given by the edges of the graph inFIG. 4. In the context of LMs, BNs are used to construct the model M as well as its parameters.

To estimate the relationship between network properties of a node I (or link), noted x_i, (e.g., hop count, rank, firmware version, etc.) and a given networking metric M_i, a linear regression may be performed. More specifically, given the following equation:

M_i=F(x_i)=b^Tx_i+ε (Eq. 2)

where x_iis a d-dimensional vector of observed data (e.g., end-node properties such as the rank, the hop count, the distance to the FAR, etc.) and M_iis the target metric (e.g., the time to join the network), which is also noted y_isometimes. Building such a model of a performance metric knowing a set of observed features is critical to perform root cause analysis, network monitoring, and configuration: for example the path delay as a function of the node rank, link quality, etc., can then be used to determine whether anomalies appear in the network and thus take some appropriate actions to fix the issue. In the equation (Eq. 2) above, the term c is a Gaussian random variable used to model the uncertainty and/or the noise on the estimate M_i. The linear regression consists in finding the weight vector b that fulfills the maximum likelihood criterion (which coincides with the least square criterion when c is Gaussian). In particular, the optimal b must minimize the Mean Squared Error (MSE):

MSE=Σ_i(b^Tx_i−y_i)²/N (Eq. 3)

where N is the total number of input data points, i.e., i=1, . . . , N.

In other words, b is a set of weights for each observed value x_i, used to compute the function F that provides the value of F. The MSE is a metric used to compute the “quality” of the model function F.

The usual approach to the solving of Eq. (2) is the ordinary least square (OLS) equation, which involves a “d×d” matrix inversion, where d is the number of dimensions. Three main problems arise immediately: (i) the dimensionality of x_imay be large, thus making OLS prohibitively expensive in terms of computational cost (approximately O(d³)), (ii) in presence of co-linearity (i.e., when several node properties are strongly correlated, as it is the case for the hop count and the ETX, for instance), OLS becomes numerically unstable (i.e., round-off and truncation errors are magnified, causing the MSE to grow exponentially), (iii) OLS being essentially non-probabilistic (i.e., it doesn't account for the whole distribution of its constituent variables, but it merely tracks averages), it cannot cope well with noise and outliers, and it is simply not applicable when ε is not Gaussian.

To overcome these limitations, the problem can be formulated as a BN (seeFIG. 5). Now, all variables are considered as random variables, even though they are all observed at this point: both input variable x_iand the output variable y_iare experimental data, and b is a (non-probabilistic) parameter of the BN at this point. By pushing this approach a little bit further, one may turn b into a random variable as well, and attempt to infer it from experimental data (that is, the observations of x_iand y_i). However, this inference problem is non-trivial, especially as one desirable feature of this learning algorithm is that it is capable of identifying non-relevant dimensionalities of x (that is, input dimensions that are weakly correlated with the output x), and automatically set the corresponding weights in b to a zero (or a very small) value.

This problem is solved by one recently proposed algorithm called Variational Bayes Least Square (VBLS) regression (Ting, D'Souza, Vijayakumar, & Schaal, 2010). Namely, this algorithm allows for efficient learning and feature selection in high-dimensional regression problems, while avoiding the use of expensive and numerically brittle matrix inversion. VBLS adds a series of non-observed random variables z_ijthat can be considered as noisy, fake targets of the factor b_j·x_ij, and whose sum Σ_jz_ijis an estimate of y_i. In turn, the weights b₃are modeled as random variables, thereby allowing for automated feature detection, i.e., the mean of b₃converges rapidly to zero if no correlation exists between the various x_ijand y_i.

VBLS estimates the distribution of the non-observed variables z_iand b using a variant of the Expectation Maximization algorithm with a variational approximation for the posterior distributions, which are not analytically tractable. Because it is a fully Bayesian approach, VBLS does not require any parameterization, except for the initial (prior) distributions of hidden parameters, which are set in an uninformative way, i.e., with very large variances that lead to flat distributions.

A Distributed and LM-Based Approach to Gather Temporary Localized Network Dynamics

As noted above, Learning Machines (LMs) are an extremely powerful tool to solve problems that involve data of high dimensionality. Although LMs may be resource intensive in terms of memory and processing power, a number of techniques have been specified in order to design distributed architectures, lightweight and still powerful algorithms hosted on constrained devices, and new approaches specifically designed for LLNs, where bandwidth among other resources are scarce. Among other challenges, LMs require to be fed with data during the learning process. Again, distributed approaches have been designed to optimize the gathering of data (e.g., thanks to compression), make use of distributed algorithms where LMs are themselves distributed thus not requiring to have one LM fed with all data, etc. Still, LLNs do not provide the ability to gather all data. This leads to incomplete datasets, making the task of learning more challenging, sometimes leading to approximations in the regression model, or noise because a specific data (e.g., feature) that could explain a behavior is lacking.

Typically in an LLN where a low-power link such as IEEE 802.15.4 is used, several thousands of nodes need to share a bandwidth of several dozens of Kbits/s, despite a number of optimizations such as load balancing among diverse paths, global scheduling, link layer enhancements such as dynamic rate adjustments, and frequency hopping. This statement is even more relevant to PLC-based network where the throughput is usually less than 10 Kbits/s, although poly-phase optimizations could help improving the throughput slightly. As a matter of fact, it has been observed in deployed networks that a number of statistics carried out by the network management protocol CoAP had to be turned off because of the extra-overhead on the network.

These ad-hoc adjustments unavoidably lead to less data and less features available to the LMs hosted on a router (e.g., Field Area Router (FAR)) or a network controller. The techniques herein, therefore, specify a generic approach to enable the triggered sending of statistical data from nodes in the network to a remote LM. The LM itself can push such triggers to the nodes, and use them to capture local and/or temporary phenomena in the network, that could advantageously be used by the LM to build its predictive model.

Said differently, in contrast with existing approaches relying on lightweight protocols such CoAP to carry metric of interest consumed by a NMS according to pre-configured specified rules, the techniques herein adopt a radically different approach whereby: 1) Requests are driven by a Learning Machine; 2) The request may specify a list of metrics of interest or computed metrics of interest that may either be specified or non-specified and linked to a specific event; 3) Reports may be sent according to a specific schedule (periodicity), a given period of Time or until explicitly cancelled by the LM.

Notably, networks have been previously configured to retrieve statistical information. Various protocol such as SNMP have been designed and deployed for decades in order to capture various network behavior metrics using GET, PUT and SNMP Traps; similarly other protocols such as IPFIX have been used to get highly granular data in a network. CoAP has been specifically designed for constrained networks such as LLNs. That being said, these protocols are all NMS/User driven: instructions to retrieve a statistical metrics are pre-determined by the user, always static and based on policies. In LLNs, CoAP has been used to report network parameters but once again, these metrics are manually configured and systematic (not directed to a specific (set of) nodes). In contrast, the techniques herein specify a generic mechanism used to gather local and temporal (ephemeral) network behavior metrics based on LM requirement.

Operationally, the techniques herein establish triggers on remote nodes in order to capture ephemeral network behavior, thereby providing additional data to train a Learning Machine (LM). For example, consider the ETX link metric, which is used by routing protocols such as RPL (and other similar routing protocols) in combination with an objective function (OF) to find the shortest constrained path. Instantaneous ETX computation by remote nodes would unavoidably lead to routing oscillation in an LLN where links characteristics do vary very quickly. Consequently, such metrics are smoothed out using low-pass filters and capture only partially the dynamic nature of the link behavior, that is, it filters away any brutal and ephemeral change. Another example is when metrics reflect instantaneous network behavior, but they are kept local to the node, simply because reporting their exact value at each change would overload the network: this is the case when a backup next-hop is used to route a packet, should the best next-hop be unreachable. In some implementation, such information, if temporary, is kept local and not reported to the FAR, NMS, network controller or the user.

Consequently, the techniques herein specify a mechanism used by an LM to effectively request the reporting of metrics characterizing temporary and/or localized network behavior in order to make the process of learning more efficient.

This type of information is critical to the accurate prediction of link or node performance (e.g., delay). For instance, both packet retransmissions, which result from transmission collisions, and queuing, which result from bursts in traffic, are causes for increased delays in LLNs. Now, an LM whose role is to predict the delay of a given node would need to have a very detailed picture of the traffic profile on the particular path to this node in order to account for these effects on the delay. Unfortunately, detailed traffic profiles are often not available, both because of the restricted bandwidth and the very limited amount of memory on the node. Instead, the LM may rely on specific traits of the traffic profile, such as the channel availability (i.e., the proportion of time there is no traffic on the channel) or peaks of traffic.

Based on these extra data, the LM may now indirectly infer the probability that a given packet encounters collisions along the path. Using the mechanisms specified herein, the LM may now request a node to start reporting channel availability if it drops below X % and traffic spikes of more than Y % than the baseline (which can be expressed in form of a threshold on the first derivative of the traffic).

The techniques herein specify a newly defined IPv6 message called the TNB( ) (Temporary Network Behavior) message that is sent by an LM to a set of nodes in a network. In contrast with existing approaches, the TNB( ) message may be directed to a set of dynamically computed nodes in the network, may or may not specify the set of metric of interest, and allows for the specification of complex rules to capture a set of metrics related to the specific events. Furthermore, the periodicity of reports may not be specified and computed by the remote nodes, which again contrasts with existing approaches.

The TNB( ) message may either be unicast (sent to a set of K nodes), multicast (sent to a set of nodes registered to a multicast group G) or broadcasted (sent to all nodes in the network); in the last two cases, in most low-power link layer, the message is broadcasted to all-nodes. The TNB( ) message is made of the following Flags and TLVs:

- Flag: Relay;
- Metric(s): Unknown, M1, M2, . . . , Mn;
- Rule: Event, Threshold, First Derivative, Second Derivative;
- Periodicity: P;
- Period of Time: T; and
- Scope (only present if the Relay flag is set to 1);

Relay Flag: when set, this indicates to the receiving nodes that the TNB( ) message should be relayed in which case the scope is specified in the Scope TLV. For example, the TSB may be sent to a node N that would relay the message with a link-local scope to it neighbor. This is an optimization in order to avoid sending the message as unicast to a set of remote nodes.

Metrics: the Metric TLV is made of sub-TLV for each metric of interest. Note that the term metric is used in a generic way and is orthogonal to the metric used by the routing protocol. The metric could be a routing metric but also any scalar or vector related to a network behavior. In its simplest form the metric is specified: for example the metric could be the RSSI of a link, the local metric computed by the node such as the second derivative of the instantaneous ETX. An alternative approach consists in not specifying the metric of interest, left to the decision of the remote node and governed by the Rule as discussed below.

Rule: the Rule TLV is used to specify the condition that will trigger the sending of a (un)specified computed metric to the LM (which may be hosted on a FAR, network controller, etc.). If the rule is a threshold, a report must be sent back to the requester if the specified metrics crosses the threshold (this is the existing mode of operation of SNMP). On the other hand, the techniques herein specify new behaviors: the rule may be the first/second derivative of a set of specified metrics, the derivative being computed locally by the node. If the rule specifies an event, then the event itself should be considered as the trigger for sending the report back to the LM for a set of (un)specified metrics.

Note that when no metric is specified, the node is expected to apply the rule for any relevant metric that is being monitored; this mechanism can be used by an LM to capture ephemeral behaviors even if it does not know the exact nature of the event it is trying to capture. As an illustration, the LM may use this mechanism to request a remote node to send a report related to locally determined metrics whose values have changed in the past X seconds. In another case, the local nodes may simply reports a large set of metrics (not knowing either which ones are relevant) that could be analyzed by the LM for relevancy evaluation.

Periodicity and Period of Time are used to indicate the requested periodicity to receive such report and the duration for which the request specified by the TNB( )applies.

A second newly defined message is used to send reports to the requesting LM using similar TLVs. As explained below, if report fusion is enabled, a newly defined F flag of the TNB( ) message is set.

For the sake of illustration, below are a few examples illustrating the mode of operation:

- The LM may request a report each time the second derivative of an instantaneously computed metric such as the ETX_Current crosses a specified threshold T1, and keep sending reports every X seconds (periodicity), potentially for a period of time T.
- If the Event=Link_failure, Metrics are next_ hop, ETX_Values, Percentage_ACK then upon detecting a link failure event the node will report the values of these metrics (a condition may be added to only report these values if and only if they cross a specified threshold).
- If the Event=New_next_hop selection, and the metrics are unspecified, this allows the remote node for determining the list of metrics that might be of interest. For example a local lightweight LM hosted on the node, may then determine a sudden changes of metrics values, a specific new event using temporal lightweight event correlation.

Cancellation of temporal metric gathering may be triggered by the LM using a specified period T or explicitly by re-issuing a TSB( ) with updated TLV values. Such an explicit notification would be triggered by the LM in the form of a control loop, should the LM have sufficient data to explain noise in its predictive model.

In yet another embodiment, the LM may schedule TSB( ) messages so as to perform fusion of reports sent back to the requesting LM (F flag set); in such as case, TSB( ) messages are sent to a set S of node, for requests related to periodic report of specified metrics, taking into account the routing topology. Indeed, the LM may advantageously select nodes sharing common paths along the routing topology for gather a set of metrics M1, . . . , Mn reported with a periodicity P and for a period of time T, thus allowing capable nodes along the path to fuse reports.

For example, if the metric of interest is related to sporadic flows between a pair of nodes, an ancestor of a set of nodes reporting local traffic flows may compress the data in the form of a lightweight traffic matrix between nodes in its sub-DAG.

FIG. 6 illustrates an example simplified procedure for a distributed and learning machine-based approach to gathering localized network dynamics. Theprocedure600 may start atstep605, continue to step610, and so forth.

Atstep610, one or more reporting nodes are selected to report network metrics in a network. Atstep615, from a monitoring node in the network, a trigger message is sent to the one or more reporting nodes. The trigger message may trigger the one or more reporting nodes to report one or more network metrics local to the respective reporting node. In response to the trigger message, atstep620, a report of the one or more network metrics is received at the monitoring node from one of the one or more reporting nodes. Theprocedure600 may illustratively end atstep625. The techniques by which the steps ofprocedure600 may be performed, as well as ancillary procedures and parameters, are described in detail above.

It should be noted that the steps shown inFIG. 6 are merely examples for illustration, and certain other steps may be included or excluded as desired. Further, while a particular order of the steps is shown, this ordering is merely illustrative, and any suitable arrangement of the steps may be utilized without departing from the scope of the embodiments herein.

The techniques described herein, therefore, provide for a distributed and LM-based approach to gather temporary localized network dynamics. In particular, the techniques herein allow an LM to capture much more detailed data about the network dynamics without unacceptable traffic overheads. This ability alone is a key enabler of advanced capabilities in learning networks where LMs are not co-located with the node they collect data from. Even in networks with large bandwidths, this type of mechanism is beneficial, as the computation of relevant features for a LM may require as much data as the network element is actually handling.

Illustratively, each of the techniques described herein may be performed by hardware, software, and/or firmware, such as in accordance with the learningmachine process248, which may contain computer executable instructions executed by the processor220 (or independent processor of interfaces210) to perform functions relating to the techniques described herein, e.g., optionally in conjunction with other processes. For example, certain aspects of the techniques herein may be treated as extensions to conventional protocols, such as the various communication protocols (e.g., routing process244), and as such, may be processed by similar components understood in the art that execute those protocols, accordingly. Also, while certain aspects of the techniques herein may be described from the perspective of a single node/device, embodiments described herein may be performed as distributed intelligence, also referred to as edge/distributed computing, such as hosting intelligence withinnodes110 of a Field Area Network in addition to or as an alternative to hosting intelligence withinservers150.

While there have been shown and described illustrative embodiments that provide for learning-machine-based predictive and proactive computer networking and associated monitoring, generally, it is to be understood that various other adaptations and modifications may be made within the spirit and scope of the embodiments herein. For example, the embodiments have been shown and described herein with relation to LLNs and related protocols. However, the embodiments in their broader sense are not as limited, and may, in fact, be used with other types of communication networks and/or protocols. In addition, while the embodiments have been shown and described with relation to learning machines in the specific context of communication networks, certain techniques and/or certain aspects of the techniques may apply to learning machines in general without the need for relation to communication networks, as will be understood by those skilled in the art.

The foregoing description has been directed to specific embodiments. It will be apparent, however, that other variations and modifications may be made to the described embodiments, with the attainment of some or all of their advantages. For instance, it is expressly contemplated that the components and/or elements described herein can be implemented as software being stored on a tangible (non-transitory) computer-readable medium (e.g., disks/CDs/RAM/EEPROM/etc.) having program instructions executing on a computer, hardware, firmware, or a combination thereof. Accordingly this description is to be taken only by way of example and not to otherwise limit the scope of the embodiments herein. Therefore, it is the object of the appended claims to cover all such variations and modifications as come within the true spirit and scope of the embodiments herein.