Rapid Spanning Tree Protocol (RSTP) supports the efficient formation and adaption of active tree topologies in a wired local area network (LAN) where multiple connections between nodes exist. Essentially, the protocol operates on a network that is a dynamic mesh of physical links to reliably form a structured tree of active forwarding links.
Traditionally, in wireless transport networks, links are planned before they are deployed. This is because topologies such as these are known ahead of time and are relatively stable, or at least changes can be controlled. An exception to this is when equipment failure occurs or a radio link become unviable (e.g., a path is blocked). To manage scenarios like these, resilience is typically designed into the topology using, for example, rings or redundant paths, and fail-over protocols are used to utilize a known (pre-planned) alternate path. Some of the protocols for Ethernet networks using wireless point-to-point links to connect bridges include Provider Backbone Bridge Traffic Engineering (PBB-TE), RSTP, and G.8032.
A different protocol is needed for wireless point-to-multipoint networks supporting dynamic topologies with relaying, multi-hop or wireless bridging. Topologies in these networks are dynamic, and nodes may be rapidly lost or discovered, making conventional network topology management inefficient.
The need exists for a system that overcomes the above problems, as well as one that provides additional benefits. Overall, the examples herein of some prior or related systems and their associated limitations are intended to be illustrative and not exclusive. Other limitations of existing or prior systems will become apparent to those of skill in the art upon reading the following Detailed Description.
Systems and methods, including procedures, are disclosed herein detailing a new approach for managing wireless point-to-multipoint networks supporting dynamic topologies with relaying, multi-hop or wireless bridging. To manage the topology of these networks, a topology management protocol can be embedded in the wireless MAC layer rather than being implemented using a hierarchical approach that detaches the transportation of packets over a link and the management of links to use to form an active network. In the case of multi-hop networks, the MAC layer controlling the links also can determine how to best utilize the available paths and what associations to form with other nodes in the network. It can because it has intimate knowledge of link status. As a result, decision making can be located within this layer for ultimate efficiency.
In summary, embedding topology management and forwarding (or bridging) into the wireless MAC layer in a multi-hop point-to-multipoint network enables self-organizing dynamic topology management by using link level information to make decisions. It also enables very efficient frame forwarding, as there is the potential for this to happen in the lower sub-layers of the wireless MAC layer where frame-by-frame decisions can be made based on link status.
The following terminology is used throughout this document and may be understood to have the following meanings. These meanings are provided for clarity and are not intended to be limiting.
Multi-hop topology management has been considered in another point-to-multipoint wireless system, the IEEE 802.16 standard through the features introduced in the 802.16j amendment. 802.16j discusses how a first node forms an association with a preferred “superordinate” node based on certain criteria. To manage topology change, the nodes operate under a protocol adapted from a general handover procedure such that the relay node or its superordinate node may trigger the handover process based on certain criteria (e.g., physical link status, capacity, latency, etc.).
This procedure is suitable for a system that supports cellular-like handover procedures and those in which links are more likely to gracefully degrade such that a handover can be affected prior to the link becoming unusable. However, systems operating in the millimeter wave bands under line-of-sight conditions using highly directional antennas can rapidly lose links. As a result, handover-like procedures in these systems would not be a good design choice.
A system according to the present disclosure establishes alternate associations so that when a link is lost, connectivity can be rapidly recovered. This system will have much lower “fail-over” time and reduced packet loss (potentially zero with a small amount of buffering), compared to a similar system that adopted handover-type approaches.
As a result, the present disclosure proposes a fresh approach to designing a self-organizing topology management protocol which can be implemented, for example, in a system with beamforming capability operating in millimeter wave bands in line-of-sight conditions that is providing resilient, low latency, and high bandwidth connectivity.
The present disclosure discusses a topology management protocol that can be embedded into a wireless MAC layer. The wireless MAC layer may be based on IEEE 802.11ad, or some other similar MAC protocol, for the purposes of supporting highly efficient layer-2 multi-hop networks to be constructed and the topology of the networks to be efficiently maintained.
Various examples of the invention will now be described. The following description provides certain specific details for a thorough understanding and enabling description of these examples. One skilled in the relevant technology will understand, however, that the invention may be practiced without many of these details. Likewise, one skilled in the relevant technology will also understand that the invention may include many other obvious features not described in detail herein. Additionally, some well-known structures or functions may not be shown or described in detail below, to avoid unnecessarily obscuring the relevant descriptions of the various examples.
The terminology used below is to be interpreted in its broadest reasonable manner, even though it is being used in conjunction with a detailed description of certain specific examples of the invention. Indeed, certain terms may even be emphasized below; however, any terminology intended to be interpreted in any restricted manner will be overtly and specifically defined as such in this Detailed Description section.
As noted above, the radio transceiver under an aspect of the invention includes an integrated antenna, in a SiP, including an AiP, format. The antenna may be configured to allow the focusing of the energy associated with a transmission in a particular direction (beam direction), such as to improve the resilience of the link in the form of an overall increase in the signal to noise plus interference ratio.
As a hub or relay station in a multi-hop point-to-multi-point or multi-hop mesh network topology, the node may incorporate 2, 3 or 4 modules combined to provide up to complete 360 deg coverage (e.g. at intersection 4 radios cover north, east, south, west directions). For a hub, if each link is at 1 Gbps, then the hub provides up to 4 Gbps capacity. Hub is the point of connection to fiber, or other backhaul mechanism behind the 60 GHz multi-hop point-to-multipoint or multi-hop mesh network.
Furthermore, the hub or relay station may be a standalone unit that is mounted on to an existing structure (e.g. a light pole, face of a building, behind a sign, at a bus or train stop, etc), or it could itself be embedded within that structure. For example a light pole could integrate 1 or more modules, and provide PoE to the module (as shown in
As an AP, basestation or endpoint station, the node may incorporate a minimum of 1 wireless communication transceiver module integrated with into a non-hub or RS station. In this case only coarse alignment is needed—point in either north, east, south or west direction towards a hub or RS. Installation can be further simplified by increasing number of modules.
One implementation shown in Plan view in
Referring back to
Using the control interface, the baseband processing engine configures the antenna beam. In some implementations, the baseband processing engine configures the antenna beam by applying a set of phase shifts to each element in the array. Alternatively or additionally, the baseband processing engine configures the antenna beam by applying a complex number that contains both phase and amplitude (gain) adjustment for each element in the array. In another form, the baseband processing engine configures the antenna beam by turning array elements on and off, this could be, for example, windows in a waveguide structure, or could be controlling polarization used for a particular baseband signal. In its simplest form, the baseband processing engine identifies a beam identifier (ID) or antenna weight vector (AWV) ID or antenna element map (AEM) ID to be used at any point in time by the RFIC. The RFIC includes a mapping of ID to actual vector or element configuration to apply, where the vector or element configuration is determined and optimized during a beam training, refinement, and/or tracking phase. In some implementations, the baseband engine provides a full AWV or AM to the RFIC, and the baseband engine maintains a list of AWVs/AEMs to use for each node that it is communicating with.
In this case, the AWV contains a set of phase and gain values where the size of the vector is equal to the number of elements in the antenna array, such that the baseband processing engine is able to control both the phase shift and any amplitude gain (or attenuation) of the signal supplied to each element (and possibly a set for each polarization of element) in the array. Or if an AEM is used, then it contains a set of settings for each element in the array (e.g. on/off), such that the baseband processing engine is able to control the effective spacing between elements in an array and/or the polarization used. Either of these approaches, and even a combination of them, allows the baseband processing engine to form various types of beam pattern and to steer both wanted energy in the desired direction, as well as to minimize the transmission of energy in the form of side-lobes, in an unwanted direction. Similarly, it allows the baseband processing engine to control where energy is received from on the receive side. It also enables a baseband engine to employ “multiple-input-multiple-output” techniques to transmit and receive simultaneously over more than one polarization to increase data rate and/or robustness. In the case where the RFIC supports simultaneous transmit and receive operations, then the baseband engine configures two sets of AWVs/AEMs to control the direction of both the transmit and receive array. In the case where the RFIC supports simultaneous transmit or receive on two different polarizations (e.g. vertical and horizontal MIMO), then there will be two AWVs/AEMs per RFIC
The MAC processing engine controls the transmission of high layer protocol (e.g. Ethernet, IP, etc.) packets over one or more wireless links between nodes implementing the disclosed architecture. The MAC engine implements software that contains algorithms and methods to facilitate communication with multiple nodes using directional antennas. It also facilitates communication with nodes not within range of the wireless link, such as by using multi-hop point-to-multipont or multi-hop techniques to communicate via other nodes. The MAC processing engine is “beam aware” and intimately involved in the control of the configuration of the radio transceiver, through the baseband processing engine, to ensure frames and packets are transmitted and received with the appropriate antenna configuration. This is achieved by supplying an associated “beam configuration” to use when transmitting the frame with each MAC protocol data unit (MPDU) that is formed and sent to the baseband processing engine that hosts the physical layer functionality. As discussed earlier, the beam configuration can be a simple index to a beam ID to use, or can be a full AWV. In some implementations, the MAC engine enables the use of a baseband processing engine that is not “beam-aware.” For example, the MAC engine may interface directly to the RFIC such that it controls the RFIC and the baseband processing engine concurrently to ensure that the signal generated (or received) by the baseband engine is transmitted (or received) with the appropriate beam configuration. In such implementations, the MAC engine would facilitate beam-forming training by configuring the baseband engine in a mode that supports this (e.g. low data rate, high processing gain) and then transitioning it to a “data-mode” (e.g. higher data rate, reduced processing gain) once training is complete. The MAC engine may generate control-frames and insert these into the data-path, e.g. in the form of specially addressed Ethernet frames, that the baseband is processing as well as control the PPF function to ensure that frames are only transferred to and from the baseband when it is operating in “data-mode”.
The MAC processing engine is also capable of supporting multi-hop point-to-multipoint or multi-hop mesh communications, or the transmission of a frame seamlessly over multiple, successive wireless links without the intervention of higher-layer protocols. It may achieve this by incorporating a layer-2 forwarding function within the MAC layer so that frame forwarding decisions can be made within the MAC layer itself, as the MAC layer is aware of the status of inbound and outbound physical links and beam settings. Incorporating a layer-2 forwarding function with the MAC layer enables rapid decision-making and optimal decisions to be made by the forwarding function that is both physical layer status and beam aware. One benefit of MAC layer relaying is that per link latency can be reduced compared to using higher-layer bridging. In addition, the utilization of the inbound and outbound physical layer link can be adjusted in harmony, resulting in more efficient transport of packets compared to where relaying was performed without context of the MAC and physical layer status. Referring to
Referring to
In general the operation and interaction between the wireless “beam-aware” MAC, baseband physical layer and RFIC enables the “beams” to become analogous to “ports” in a wired layer-2 Ethernet switch, with the wireless MAC layer managing the efficient forwarding of frames from one “port” (which is actually a “beam”) to another.
In addition, the functions of the MAC layer support auto discovery of other nodes and maintenance of wireless links found to other nodes without user intervention. The MAC layer also supports the configuration of appropriate frame and packet forwarding or configuration of tunnels to aid forwarding of frame and packets over multiple wireless links between the source and destination node in the wireless network. Auto discovery is supported by nodes that are established and operating in the network transmitting “beacons” or signals that identify their presence, such that nodes wishing to associate with the next node can learn of nodes to which they can gain access to the network. As the beamforming is used by the transmitter, this involves having nodes attached to the network “beam sweeping” the transmission of the beacon. This is achieved by transmitting the beacon multiple times, over a period of time, each time sending it to the baseband processing engine for transmission using a different beam ID (or AWV). The receiver typically listens for such transmissions using either a quasi-omnidirectional receive mode, or some form of coarse antenna beam to enable it to hear the beacon. As full optimal receive side beamforming is typically not available, the beacon is sent using very robust transmission approaches so that it can be received by a node that is not implementing high gain receive side beamforming. Once the beacon is received and the transmitter and receiver are essentially synchronized, then receiver side beamforming can take place to enable the transmitter and receiver to communicate using more spectrally efficient encoding schemes thereby achieving the target throughput rates.
The layer-2 PPF enables higher layer packets sourced from, or destined to, a wireline network to be transported over the wireless link. At a minimum the PPF is responsible for translating Ethernet frames into wireless MAC frames. In the case there are no frames to be sourced from or supplied to the wireless network, the PPF may at a minimum act as a power source using power-over-Ethernet technologies. In the case of a node with multiple modules, then the PPF on one of the modules may provide a “master” function, which may include acting as a layer-2 bridge or switch. Referring to
The system effectively converts Ethernet frames (and packets encapsulated within them, such as IP and/or MPLS) to mmWave transmission, and performs the reverse process for receiving frames and packets to/from multiple sources. The module is powered either separately or by using a shared Ethernet and power interface, commonly referred to as PoE (power-over-Ethernet).
Referring to
Full duplex relay with full duplex links: Module 1 can be communicating with an AP, while Module 2 is relaying frames to/from the hub;
Full duplex relay with half duplex links: Module 1 can be receiving from an AP while transmitting to the hub (or transmitting to the AP while receiving from the hub);
Half-duplex relay with full duplex links: Module 1 can be communicating with the AP at one point in time, then relaying frames to the hub at another point in time; and
Half duplex relay with half duplex links: Module 1 is either transmitting or receiving to/from the AP or node, alternating in time between direction and transceiver function.
As well as supporting half-duplex operation with time division duplexing, two arrays can support full duplex link operation, or full duplex relay/half-duplex link operation using frequency division duplexing. Full duplex link operation is achieved by a transmit (Tx) and receive (Rx) array pointing to the same node and allowing packets to be transmitted and sent at the same time. Frequency division full-duplex (FDD) operation is supported without the need for analog domain channel or sub-band filter, commonly referred to as a duplexer or diplexer, and typically required in any FDD communication system that has to share certain elements of the transmitter or receiver (e.g. antenna). Due to the high-level of integration proposed and the operating frequency, it is possible to ensure sufficient isolation between the transmitter and receiver components and antenna arrays to prevent the transmitted signal interfering with the received signal in the analog domain. In light of this, transmitted signals may be filtered from received signals entirely in the digital domain.
Isolation between transmitter and receiver components is achieved in various ways. Isolation may be achieved by using separate transmit and receive antennas that are physically separated. At mmWave frequencies the separation does not need to be large due to the short wavelength. Isolation may also be achieved using beamforming to ensure that both the transmit and receive arrays are focused away from each other. Isolation may further be achieved by using robust modulation and coding schemes, which can be used due to the abundance of bandwidth, meaning that any residual leakage of energy after processing in the digital domain has minimal impact on receiver performance. For example, separation of a few centimeters provides at least 30 dB of isolation; in addition, with beamforming applied, the transmit and receive sidelobes can be ˜30 dB attenuated. The net result is a combined analog domain isolation of >90 dB which is of the order of that provided by a traditional duplexer. Further isolation could be provided by building low-profile “wall” (e.g. a sufficiently designed metallic, or other material, insulator) between the two arrays to reduce the effective coupling of signal between the two arrays. It is possible that as well as enabling improved operation (e.g. at higher order modulation and coding schemes, or reduced digital domain processing requirements) on adjacent frequencies, that with sufficient additional attenuation by a wall that isolations of >100 dB could be achieved enabling full duplex operation where the same channel is used for both transmit and receive.
Full duplex relay operation is achieved by the Tx array pointing to one node while the Rx array points to the other node; the Tx/Rx arrays then alternate over time to allow relaying of frames in both directions. This mode of operation can be particularly beneficial in networks with highly asymmetric traffic: e.g. downlink centric where data is generally flowing from hub to relay to AP to end-point. It also allows a module pointing in a coarse direction that needs to perform relay function to operate efficiently (e.g. hub and an AP are both North of the RS). Alternatively if only one array is available, or only one array can be active at any one point in time, then half-duplex operation can be supported.
In the general sense, due to the lack of an analog duplexer or diplexer, the frequency channel used for transmit and that used for receive can be defined in software, as well as whether the system is operating in full or half duplex, with frequency or time division duplexing, such that software-defined duplexing (SDD) is enabled.
Overall, the functional blocks shown in
The functional blocks in
Overall, the functional blocks shown in
The functional blocks in
The system supports centralized operations and maintenance (OAM) and facilitates the node and architecture to be self-organizing, in the sense that the network of nodes will be a dynamic self-organizing network (SON) supporting multi-hop point-to-multipoint or multi-hop mesh topologies. To facilitate this software defined networking (SDN) approaches may be utilized, including the use of OpenFlow, such that some of the control plane functionality required to support the operation of the node in a network of nodes is provided by a centralized controller. In this architecture each node presents an application programming interface (API) to allow the centralized function to control the behavior of the node within the network of nodes.
The system described above includes several elements, combined together, to create a new type of wireless communications system (hardware and embedded software) that is able to provide low latency, Gbps communications over much longer ranges than may otherwise be possible. In addition, the approach of using centralized OAM, dynamic SON and SDN (and SDD in the case frequency division duplexing is required) enables a large network of numerous nodes to be deployed and operated with ease, and for the network to be able to self-optimize based on traffic patterns and changes in topology caused when certain wireless links become available or unavailable between any two nodes within the network.
The system can be realized using a number of system-on-chip (SoC) and system in package (SiP) devices (integrated circuits and systems) mounted on to a printed circuit board (PCB). Alternatively the various elements can be implemented on separate silicon dies and integrated into one or more SoCs or SiPs, and ultimately all the elements can be implemented on a single silicon die and packaged in a SiP.
The hardware module described above forms a basic building block that has multiple features, including: capable of being combined with access nodes either as an integral module, or as a field pluggable device, to provide metro-wide transport connectivity; capable of being packaged with one or more other modules to provide a “relay” function to allow two or more access nodes to connect; capable of being packaged with other modules to provide a “hub” function to allow nodes to connect to a fiber link to the core network; etc.
A node with multiple modules may incorporate all of the elements described on each module, or one of the modules can behave as a master module, itself driving, for example, just the PHY and/or RF element on one or more other modules.
One benefit of such a solution is that it enables a number of very low cost, high capacity simple wireless links to be provided, leveraging highly integrated and relatively low-cost electronics, but, by relying on intelligence in the software residing on the modules, enables dynamic, adaptive, low latency and resilient multi-hop point-to-multipoint or multi-hop mesh networks to be formed. As such it enables the potential to offer a much lower cost per GB solution, but also enables the easy deployment of a very resilient network.
One of ordinary skill in the relevant art will recognize that, although not required, aspects of the invention may be implemented as computer-executable instructions, such as routines executed by a general-purpose data processing device, e.g., a server computer, wireless device, personal computer, etc. Those skilled in the relevant art will appreciate that aspects of the invention can be practiced with other communications, data processing, or computer system configurations, including: Internet appliances, hand-held devices (including personal digital assistants (PDAs)), wearable computers, all manner of cellular or mobile phones (including Voice over IP (VoIP) phones), dumb terminals, media players, gaming devices, multi-processor systems, microprocessor-based or programmable consumer electronics, set-top boxes, network PCs, mini-computers, mainframe computers, and the like. Indeed, the terms “computer,” “server,” and the like are generally used interchangeably herein, and refer to any of the above devices and systems, as well as any data processor.
Aspects of the invention can be embodied in a special purpose computer or data processor that is specifically programmed, configured, or constructed to perform one or more of the computer-executable instructions explained in detail herein. While aspects of the invention, such as certain functions, are described as being performed exclusively on a single device, the invention can also be practiced in distributed environments where functions or modules are shared among disparate processing devices, which are linked through a communications network, such as a Local Area Network (LAN), Wide Area Network (WAN), or the Internet. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.
Aspects of the invention may be stored or distributed on tangible computer-readable media, including magnetically or optically readable computer discs, hard-wired or preprogrammed chips (e.g., EEPROM semiconductor chips), nanotechnology memory, biological memory, or other data storage media. Alternatively, computer implemented instructions, data structures, screen displays, and other data under aspects of the invention may be distributed over the Internet or over other networks (including wireless networks), on a propagated signal on a propagation medium (e.g., an electromagnetic wave(s), a sound wave, etc.) over a period of time, or they may be provided on any analog or digital network (packet switched, circuit switched, or other scheme).
The system may work with various telecommunications elements, include 2G/3G/4G network elements (including base stations, Node Bs, eNode Bs, etc.), picocells, etc. Alternatively or additionally, the network includes an IP-based network that includes, e.g., a VoIP broadcast architecture, UMA or GAN (Generic Access Network) broadcast architecture, or a femtocell broadcast architecture. (Unlicensed Mobile Access or UMA, is the commercial name of the 3GPP Generic Access Network or GAN standard). Of course, VoIP using WiFi access points (APs) or other nodes of an IEEE 802.11 network may be used.
This section discusses high-level methods and procedures for managing topology formation, adaption and frame forwarding. In some implementations, the methods and procedures are used in a wireless network in which at least some nodes are point-to-multipoint capable, as well as capable of forwarding or relaying (wireless bridging) frames. It is anticipated that these procedures will form the basis for a topology management protocol that could form an extension to an existing set of MAC protocols, such as those defined in IEEE 802.11ad, or as part of a proprietary MAC protocol.
In some implementations, the methods and procedures are used in a system communicating in the 60 GHz band and relying on dynamic beamforming in order to support point-to-multipoint communications over reasonable link distances. The system is also being used to effectively extend the service of a wider area network out to end-nodes.
One of the key differences between topology management for dynamic wireless networks and wired Ethernet with RSTP is the notion of association available in wireless networks versus the control over the formation of physical layer links that is available in wired Ethernet networks. For a variety of reasons, nodes in a wireless network form associations with each other before an active communications path is formed. The result is that a wireless multi-hop point-to-multipoint network will naturally form as a loop-less tree topology, as it has control over what physical layer connections to establish.
However, a protocol is needed to ensure the efficient formation of an initial tree when the network starts up and to provide support for dynamic adaptation of the topology, in a self-organizing manner, when a node joins or leaves the network, or when a link is lost or becomes available.
In general, after it is initially formed, three events can a change in topology:
Any of these events will cause a topology change that could effectively ripple through the network and cause other nodes to change their points of primary association. Therefore, any of these events may trigger wide-scale topology adaption throughout nodes of a network.
Prior to considering these events that cause changes in topology, the first subsection below describes how the initial network topology is formed. The second subsection then addresses events that may cause an adaption in topology.
In some implementations, in a wireless network, like those described herein a node does not advertise itself as being available for other nodes to connect to (i.e. it does not broadcast a beacon indicating its existence) until it has itself connected to a network and can provide connectivity to a WAN, unless it can itself provide some higher layer service to other connecting nodes.
The first station to start broadcasting its availability is therefore a node with connectivity to the WAN, or ability to provide some higher layer service to other nodes, this type of node is called a hub node and it is configured as such by the fact that it knows a connection exists to the WAN (e.g., fiber, wired Ethernet or other wireless connection), or it can provide higher layer services (e.g., access to stored data or other device that stores or generates data that is to be made available). This node is similar to the root bridge in the case of RSTP.
All non-hub nodes are initially scanning to find a node to which they can attach. Nodes that receive beacon information from a hub node consider associating with the hub node. Once a non-hub node has associated with a hub node, the non-hub can also advertise its existence to enable other non-hub nodes, such as those out of communication range with the hub node, to discover and connect to them.
Nodes broadcasting their existence also broadcast other information, including a “hub path cost.” A node detecting multiple nodes broadcasting their existences can evaluate hub path costs received from each node to determine the best node to associate with. In some implementations, a primary association is formed with the node broadcasting the best hub path cost.
Hub path cost is a metric quantifying a cost for accessing a hub through a node. In some implementations, hub path cost is a function of the characteristics of the path between the node advertising the hub path cost and the hub node. In some implementations, characteristics include a total number of links, or hops, that constitute a hub path. For example, if a hub path consists of 4 links, the hub path cost would be 4. In some implementations, characteristics include an aggregate latency over the links that constitute the hub path. For example, if each link of a hub path adds 1 millisecond of latency, the hub path cost may be represented as 4 milliseconds. In some implementations, characteristics include a cost associated with accessing a WAN via a hub node. For example, a hub path cost indicative of a latency of a hub path may factor in a latency associated with a hub node. Accordingly, the hub path cost may be representative of a latency between the node and a WAN. In some implementations, a hub path cost is determined by the amount of bandwidth available over the path provided by the node advertising the hub path cost, which in itself would be the minimum of the hub path cost advertised by the next upstream node and that available to any connecting node. As can be appreciated, a hub path cost may indicate at least one of latency and available capacity, which are useful for a connecting node to evaluate when selecting the best point of attachment. In the case of a latency related metric for hub path cost, a lower value represents a better hub path cost, whereas in the case of available bandwidth, a higher value represents a better hub path cost. In some implementations, a hub path cost is calculated using an inverse function of available bandwidth, or “path utilization” where path utilization could be an absolute value (e.g. average Mbps or Gbps utilization) or relative (e.g. average 10%, 50%, 75% of available bandwidth used). Adopting this approach would mean that irrespective of if latency or utilization is used in derivation of the hub path cost, the principle that lower is better will always apply.
In some implementations, hub path cost is calculated based on a composite of latency and utilization (or available bandwidth), indicated either as two separate values or weighted to calculate a single hub path cost value. If a weighted value is used, the formation of the weighting function may be arranged such that a high value represents high latency, high utilization (or low available bandwidth), or a combination of both. Conversely, a low value would indicate low latency and low utilization (or high available bandwidth). With this approach for calculating hub path cost, a higher value indicates a worse hub path than a node that advertises a lower value.
Hub nodes can also broadcast a hub path cost. A hub path cost associated with a hub node reflects a cost for accessing a WAN or service via a direct connection to that hub node. In some implementations, a hub path cost associated with a hub node is indicative of a current latency to access a WAN or service through the hub. In some implementations, hub path cost for a hub node reflects a status of the hub node, such as its available capacity, its utilization (e.g., a percent of its bandwidth that is utilized), a typical access latency for the hub node, or the like. In calculating a hub path cost, a node that is not a hub node may factor in a hub path cost of a hub node that it is connected through to a WAN. In some implementations, the hub node is itself a server computer offering services, and hub path cost reflects a cost for accessing those services.
Hub path cost can also consider other factors or be derived using other methods. In a general sense, a hub path cost is a representation of one or more key path related metrics that provide a simple value by which a node seeking to join a network may determine which is the “better” path to use; in addition in the general sense the hub path cost could be a single value derived from a single metric, a single value derived from a number of metrics by applying some weighting function, or multiple values each representing a single metric and/or the result of a weighted function applied to two or more metrics.
A node may also form alternate associations in addition to its primary association. Under a procedure for establishing alternate associations, the node performs an association procedure but indicates that the alternate association is an alternate association for the node, instead of a primary association, that is being setup for purposes of rapid failover.
As a general summary, an approach of topology formation includes the following steps:
Under the procedure of
Once a topology is formed, support for adaptation may be needed for at least the following cases to enable the network to continue to be self-organizing:
The remainder of this section describes procedures for addressing at least these scenarios.
2.2.1 Node Addition
A node joining the network may follow the same association procedure described above and depicted in the example from
After a new node joins the network, the active network topology may be suboptimal. For example, a node already having a primary association through another existing node may be able to have a better primary association through the new joining node. That is, the hub path cost through the new node may be less, or better, than the hub path cost through the node's existing primary association. However, until the existing node detects and assesses the hub path cost of the new node, it will be unaware of the better point of association.
This scenario is particularly problematic in a network that inherently relies on beamforming to enable nodes to communicate with each other, as an existing node may not have resources (e.g., time) to dedicate to blind scanning to detect new nodes. Even if the existing node did have resources, it could take some time before the existing node is able to detect the existence of a new node, depending on how much time it can devote to blind scanning. Also, in a network where changes in topology might not be that frequent, but rapid reaction is required when they do occur, continual periodic scanning would not be an efficient use of resources.
A protocol according to the present disclosure may limit a node from blind scanning once it is associated and active, but allow efficient adaption in topology to enable it to exploit the potential to efficiently optimize its point of primary association.
In some implementations, a protocol according to the present disclosure may employ at least one of the following approaches to manage the discovery and evaluation of a new node:
The first of these approaches is essentially passive discovery of a topology change, by scanning rather than being informed. As described above, this is suboptimal in the case of a millimeter wave system utilizing beamforming, as this could result in either lots of time spent scanning, and therefore lost time for supporting active communications, or latency in time for a network to self-optimize when a new node joins the network.
The second involves the node sending a broadcast frame out to other nodes in the network, once associated, using its primary association and relying on the receiving node to propagate this frame over the network to inform existing nodes that there is a new node in the network. This message may contain information pertinent to assisting existing nodes to evaluating this new node as a potential point of primary association (e.g. hub path cost, frequency of operation, time of broadcast beacon signal, waveform encoding information, geographic location, etc).
Message propagation needs to be managed to prevent unnecessary wide-scale flooding in the case of a large network, as it would be pointless to inform nodes several 100 s of meters away of a new 60 GHz radio node, which the distant nodes would not be able to directly communicate with over the air, due to propagation loss being too great. In some implementations, a “lifetime” (e.g. maximum hop count) can be associated with a broadcast message, with the count decremented each time the message is forwarded. Once the count reaches zero the message is discarded and does not propagate further through the network. While this is a simple approach, there is the potential for inefficiencies due to messages propagating further into the network than required, triggering nodes to scan that have zero chance of detecting the messaging node (thus wasting resources), or messages not getting to nodes due to life-time expiration that could detect the new node.
The third approach is often the most efficient and robust. Under this approach, when a new node is evaluating nodes for primary association it effectively discovers all nodes within communication range, i.e., all nodes that will be able to detect it once it starts transmitting its beacon. As in other approaches, the new node selects its primary association and calculates its hub path cost. Then, instead of advertising its existence by sending a message on the primary association, it communicates directly over the air with other nodes that it previously discovered in order to form reciprocal alternate associations. After doing so, a node receiving the alternate association proposal can evaluate if the new node can provide a better hub path cost than its existing primary association. If it can, that receiving node can switch its primary association to be to the node joining the network, instead of forming an alternate association with it. If not, then the association becomes alternate for both nodes and could be used as a primary association in the future by either node, if the current primary association fails.
Under the procedure of
Under the procedure of
One substantial benefit is that this method and set of procedures enable a new node to join the network, and for existing nodes to act to optimize the topology in a distributed, dynamic manner without planning or centralized control.
While nodes may exchange frames over alternate associations for maintaining the association, no frames will be forwarded using the alternate association.
2.2.2 Controlled Node Exit
A node can leave a network either gracefully (controlled) and also abruptly (uncontrolled).
In a controlled case, such as when an operator initiates the power down of a node or a known fault is about to occur (e.g., over temperature, soft failure, etc.), the node can exit the network informing other nodes that it is going offline. In an uncontrolled case, the node cannot inform other nodes of its exit from the network, and those other nodes therefore have to detect that the exiting node is no longer available or operational.
In both cases the topology may need to adapt to heal lost primary associations. In the former case this can happen in an efficient controlled manner; in the latter case methods are needed to ensure rapid recovery and to minimize frame loss.
2.2.2.1 Controlled Node Exit
Two operations may be required when a node makes a controlled exit from a network. A first is to initiate disassociation of any nodes that have primary associations with the node exiting the network. A second is to disassociate the exiting node's primary association. In some implementations, there may be some delay between the two operations, to ensure that any frames that happen to arrive in the interim are forwarded on.
When a node intends to exit the network, it first sends a disassociation request message to the nodes that it provides the primary point of association to. The receiving nodes may respond to indicate to the exiting node whether the node exit will cause the receiving node to be orphaned (i.e., there are no alternate associations that can become primary). When a node would be orphaned, the exiting node (or operator of it) may reevaluate, depending on the reason for exiting the network.
The next step is for the nodes to respond back to indicate they are ready to switch the primary association, and an alternative hub path is ready. While waiting for the confirmation of disassociation the exiting node may continue to forward any received frames.
Once all nodes with a primary association to the exiting node have confirmed establishment of the new hub path, the exiting node can confirm its intent to exit in order to stop associated nodes from transmitting any more frames through the exiting node. It can then continue to exit the network by conducting the second procedure of disassociating its own primary association. Once this is complete the node sends a final message to the associated nodes to release the radio link, as no more frames will be pending for forwarding to them.
2.2.2.2 Uncontrolled Node Exit
In the event of an uncontrolled exit by a node, because of, for example, a catastrophic failure (e.g. loss of power, equipment failure), the exiting node may not be able to inform nodes with primary associations to it, or the node with which it has its primary association. Associated nodes may rapidly detect a loss of their primary association by identifying any of a number of events (e.g., lack of response to requests, lack of timely acknowledgements, loss of beacon, etc.). All of these can indicate radio link failure (RLF) status in the MAC, triggering fail over for an associated node to an alternate association.
To minimize frame loss, the node providing the exiting node's primary association can buffer unacknowledged transmitted frames. Once the forwarding tables are updated and the new path is known to the destination address, then buffered frames can be forwarded on to the destination using the new path. Methods for forwarding table update following an uncontrolled node exit are described in Section 2.2.4.2.
2.2.3 Point of Primary Association Change
Point of primary association change can occur due to any of a number of events, including:
The first and second events have been discussed above. The third event can occur due to a number of scenarios that are described further in this subsection.
The fourth event is may be treated in the same way as for the uncontrolled node exit scenario discussed above, except that a single point of primary association is lost. The node to which the association is lost may act in the same manner as the node that hosts the primary association of an exiting node, buffering and then forwarding frames to minimize frame loss during the period that the node that lost its primary association reconfigures its primary association.
The remainder of this subsection describes procedures for the third scenario, which can be caused, for example, by the following:
Each case is considered in further detail in the following subsections. In all cases, when a situation occurs that causes a better hub path cost to be detected by a node, then the node will change its point of primary association. It does this by either making a new association and demoting its current primary association to an alternate association, or by switching an alternate association having a better hub path cost to primary and demoting its current primary to alternate.
2.2.3.1 Detection of New Link
A new link may be detected by a node through periodic scanning using spare time intervals. In some implementations, detection of new links is assisted by information from neighboring nodes, which provide a summary of the nodes in the neighborhood and possibly information to assist in efficient detection of their beacons. The neighborhood information could be obtained by multiple means, such as through distributed approaches involving probing of nodes by requesting the node with the primary association to distribute probing messages over the neighborhood, or by collecting information in a central server that can be queried by any one node.
Once a node is in receipt of neighborhood information, it can use this to optimize a proactive attempt to detect neighborhood nodes that it does not have a primary or alternate association with. By using a proactive approach for detecting neighboring nodes, a node may identify a previously undiscovered node that has a better hub path cost than a node with which it has a primary or alternate association. For example, when joining a network, a node might only detect one other node and it may be associated with an undesirable hub path cost. Given the limited options available, the node may form a primary association with this detected node. However, a change in environment may cause an existing node to become viable. For example, rain or an object that was obscuring a link between the node and the existing node may disappear. By proactively searching for new links to existing nodes, the node can more readily find a more desirable node with which to form an association.
In summary, one method, as hinted to above, involves querying a centralized database (e.g. centralized SON server) for information about nodes in the vicinity, and neighborhood information could include location information, frequency of operation, time of beacon, etc—all of which would assist in attempting to discover an existing node.
In some implementations, a distributed approach is used, under which a node “probes” for other nodes in the neighborhood. A simple approach is to define a neighbor probe message with a limited lifespan that propagates out over the network via the primary association, with the message “hop” distance limited to prevent full network-wide propagation. This message would likely include location information and a request for all nodes receiving the message within a given distance (e.g. 100 m) to respond back with information regarding its hub path cost, beacon transmission information, and so forth. This would assist the receiving node in identifying nodes worthy of attempting to discover (e.g. better hub path cost, good hub path cost for an alternate association, etc.) and in efficiently discovering those nodes using a directed rather than blind scan.
In some implementations, a node advertises known information about a neighborhood in a beacon frame, or similar, instead of, or in addition to, using a query based mechanism.
2.2.3.2 New Network Node
The second case is when a new node joins the network, which is described in Section 2.2.1 when the node receives an alternate association proposal having a hub path cost better than the hub path cost associated with the node's current primary association.
2.2.3.3. Upstream Topology Change
The third case is caused when, due to of a change in network topology, a hub path cost associated with an alternate association is better than a hub path cost associated with a node's primary association. This may be discovered by the node through periodic monitoring of the hub path costs of the primary and alternate nodes. In this case, the node reacts by switching the point of primary association, swapping the primary and alternate states of the two associations.
2.2.4 Forwarding Table Update
After a node changes its point of primary association, the path to the hub will change and forwarding tables at intermediate nodes on the old and new path will need to be updated to remove stale entries and to ensure appropriate new entries are added.
In RSTP, a topology change notification message is used to propagate an event like this through a network. In some implementations of the present disclosure, nodes adopt a similar approach.
In some implementations, this approach is used in the case of a controlled change in the point of primary association. In some implementations, in the case of uncontrolled node exit, it is impossible for the exiting node to communicate a topology change notification and, instead, the nodes associated with the lost node can send the topology change notification. These two different cases are described in the two following subsections.
2.2.4.1 Controlled Topology Change
Once the point of primary association is changed, the node can then broadcast out a topology change notification message. In some implementations, nodes that receive this message on a connection that is not their primary association do three actions:
In some implementations, a node that receives the message on its primary association does the following three actions:
Under the procedure of
As with the neighborhood probe message described earlier, the topology change notification could be a complete network wide flood, or it could have a limited hop count to limit how far such a message will propagate through the network. In reality, it only needs to be received by the nodes that formed the previous hub path, therefore the best approach is for the message to only ever propagate in the direction of the hub until an entry for the node in question is found in a forwarding table. At this point the message can then be forwarded down the old hub path to remove stale entries. Adopting this approach ensures that the topology change notification is only propagated to nodes that need to receive the information.
2.2.4.2 Uncontrolled Topology Change
In the uncontrolled case where the node exits the network, the node that hosted the primary point of association can detect an uncontrolled exit and use the topology change notification to inform other nodes on the hub path that an exit has occurred. This enables them to remove stale information from forwarding tables. This message may be time-stamped, so that should the node that exited manage to recover association to the network through another node, and the new hub path converges with the old hub path at some point, that delayed propagation of the topology change notification with uncontrolled exit information does not cause a new forwarding table entry to be subsequently deleted.
This enables a node receiving the notification with uncontrolled exit information to determine if a new forwarding table entry has occurred after the uncontrolled exit, so that it can determine if the information is current or stale, and hence persists or is removed.
2.2.5 Ensuring Topology Stability
To ensure topology stability, in some implementations, appropriately configured timers are utilized in a number of cases before certain procedures are triggered. Some examples include:
Nodes may also monitor a particular primary association and if it is continually changing between primary and alternate, causing frequent topology changes, it may choose not to make the node with the best hub path cost the point of primary association in order to improve stability in the network. Or it may extend the duration of the timer used between detection of a node with a better hub path cost value and the point at which it acts to change the point of primary association to this node.
Unless the context clearly requires otherwise, throughout the description and the claims, the words “comprise,” “comprising,” and the like are to be construed in an inclusive sense, as opposed to an exclusive or exhaustive sense; that is to say, in the sense of “including, but not limited to.” As used herein, the terms “connected,” “coupled,” or any variant thereof means any connection or coupling, either direct or indirect, between two or more elements; the coupling or connection between the elements can be physical, logical, or a combination thereof. Additionally, the words “herein,” “above,” “below,” and words of similar import, when used in this application, refer to this application as a whole and not to any particular portions of this application. Where the context permits, words in the above Detailed Description using the singular or plural number may also include the plural or singular number respectively. The word “or,” in reference to a list of two or more items, covers all of the following interpretations of the word: any of the items in the list, all of the items in the list, and any combination of the items in the list.
The above Detailed Description of examples of the invention is not intended to be exhaustive or to limit the invention to the precise form disclosed above. While specific examples for the invention are described above for illustrative purposes, various equivalent modifications are possible within the scope of the invention, as those skilled in the relevant art will recognize. For example, while processes or blocks are presented in a given order, alternative implementations may perform routines having steps, or employ systems having blocks, in a different order, and some processes or blocks may be deleted, moved, added, subdivided, combined, and/or modified to provide alternative or subcombinations. Each of these processes or blocks may be implemented in a variety of different ways. Also, while processes or blocks are at times shown as being performed in series, these processes or blocks may instead be performed or implemented in parallel, or may be performed at different times. Further any specific numbers noted herein are only examples: alternative implementations may employ differing values or ranges.
The teachings of the invention provided herein can be applied to other systems, not necessarily the system described above. The elements and acts of the various examples described above can be combined to provide further implementations of the invention. Some alternative implementations of the invention may include not only additional elements to those implementations noted above, but also may include fewer elements.
Any patents and applications and other references noted above, including any that may be listed in accompanying filing papers, are incorporated herein by reference. Aspects of the invention can be modified, if necessary, to employ the systems, functions, and concepts of the various references described above to provide yet further implementations of the invention.
These and other changes can be made to the invention in light of the above Detailed Description. While the above description describes certain examples of the invention, and describes the best mode contemplated, no matter how detailed the above appears in text, the invention can be practiced in many ways. Details of the system may vary considerably in its specific implementation, while still being encompassed by the invention disclosed herein. As noted above, particular terminology used when describing certain features or aspects of the invention should not be taken to imply that the terminology is being redefined herein to be restricted to any specific characteristics, features, or aspects of the invention with which that terminology is associated. In general, the terms used in the following claims should not be construed to limit the invention to the specific examples disclosed in the specification, unless the above Detailed Description section explicitly defines such terms. Accordingly, the actual scope of the invention encompasses not only the disclosed examples, but also all equivalent ways of practicing or implementing the invention under the claims.
The present application claims priority to and the benefit of U.S. Provisional Patent Application No. 61/841,170, filed Jun. 28, 2013, entitled “SELF-ORGANIZING TOPOLOGY MANAGEMENT PROCEDURES, IMPLEMENTED IN FOR, FOR EXAMPLE, MILLIMETER-WAVE MULTI-HOP POINT-TO-MULTIPOINT NETWORK NODES.” The disclosure of the above-listed application is hereby incorporated by reference in its entirety.
Number | Name | Date | Kind |
---|---|---|---|
5870605 | Bracho et al. | Feb 1999 | A |
7756041 | Whitehill et al. | Jul 2010 | B2 |
8103789 | Gan | Jan 2012 | B1 |
20060095815 | Noy | May 2006 | A1 |
20090201844 | Bhatti | Aug 2009 | A1 |
Number | Date | Country | |
---|---|---|---|
61841170 | Jun 2013 | US |