System and methods for synchronizing edge devices on channels without carrier sense

Abstract

A system and method are disclosed for synchronizing edge devices on a network, wherein the network has limited bandwidth and does not support a practical carrier sense mechanism. The edge devices transmit, using a slotted protocol, to a server located at a central data aggregation point. The server also controls a Central Transmitter, which sends messages to the edge devices to assign transmission slots and provide timing information to the edge devices to ensure that transmissions sent by devices sharing the same communication channel do not collide.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention is directed generally toward the domain of network resource management where the message bearing substrate for a multiplicity of transmitters sited at or near the edge of the network has limited bandwidth and does not support a practical carrier sense mechanism. An example of such a network is the electrical distribution grid. In such a network, if two or more transmitted messages overlap in time (collide) on the same bearer channel, neither message is decipherable at its destination. Slotted protocols, where every transmitter is allocated specific times wherein it is permitted to transmit, can maximize the utilization of such network. However, collisions are prevented only if the time clocks on every transmitter are precisely synchronized with one another. It is understood that “precisely” means to have a known tolerance as opposed to “exactly.” The present invention describes a system and diverse methods for synchronizing endpoints on such a network.

2. Background of the Invention

The electrical grid in the United States and most other areas of the world is historically divided into two networks: the transmission grid and the distribution grid. The transmission grid originates at a generation point, such as a coal-burning or atomic power plant, or a hydroelectric generator at a dam. Power is generated, and transmitted to distribution points, called distribution substations, via a highly controlled and regulated, redundant, and thoroughly instrumented high-voltage network which has at its edge a collection of distribution substations. Over the last century, as the use of electrical power became more ubiquitous and more essential, and as a complex market in the trading and sharing of electrical power emerged, the technology of the transmission grid largely kept pace with the technological requirements of the market.

The second network, the distribution grid, is the portion of the electrical grid that originates at the distribution substations and has at its edge a collection of residential, commercial, and industrial consumers of energy. In contrast to the transmission grid, the technology of the distribution grid has remained relatively static since the mid-1930s until very recent years. Today, as concern grows over the environmental effects of fossil fuel usage and the depletion of non-renewable energy sources, interest has increased in augmenting the electrical distribution grid with communication instruments. The primary goals of this activity are energy-related goals such as energy conservation, resource conservation, cost containment, and continuity of service.

It is desirable when so augmenting the distribution grid with communication instruments to employ the grid itself as the communication medium. The medium-voltage (>1 KV) power lines that comprise the bulk of the distribution grid discourage tampering, snooping, and spoofing. Public utilities in the United States are subject to a statutory accounting model that favors capital investment over facilities costs, and so would rather communicate via the grid, a network they own, than the Internet or other wide-area wired or wireless network infrastructure, which they must lease or for which they pay usage charges. Finally, using the grid as the communication medium enables a mechanism that allows the network to infer the schematic location of the transmitter on the grid from characteristics of the received signal at another location on the grid. This capability has high value to utilities for applications such as fault isolation, energy optimization, and recovery management. Finally, the electrical grid is (reflexively) always present where instrumentation of the grid is needed, whereas nearly all utilities encounter situations where conditions do not support the use of other types of networks, or where the cost of other network types is prohibitive.

Some characteristics of the electrical grid are greatly different from other networks. First, as in many wireless applications, but not in other wired applications such as Ethernet, there is no carrier sense. In other words, one transmitter will generally not be able to know if another transmitter is transmitting on the same channel it wishes to use. If two transmitters attempt to use the same channel simultaneously, neither signal will be correctly received. Second, the data rate achievable using the power grid as a transmission medium is low compared to modern fiber or wireless networks. Finally, the network is hierarchical in structure and asymmetrical in transmission cost. It is much easier and cheaper to transmit from low voltage to high voltage on the electrical distribution grid than from high voltage to low voltage.

Despite these apparently limiting characteristics of the medium, the high value of the enabled applications, especially inferring the schematic location of the transmitter, drives the development of systems and methods for maximizing the utilization of a network with these characteristics. While the desire to use the electrical distribution grid as the network substrate is described in one aspect of the present invention, it must be noted that other network media have similar characteristics. There is nothing about the present invention that limits its application to only the electrical distribution grid.

Most conventional networks take advantage of the stochastic nature of transmission patterns. Transmissions from the edge of the network are initiated unpredictably and infrequently, from any specific transmitter, with respect to the data rate of which the conventional network is capable. The actual utilization of the network, expressed as a percentage of the maximum achievable data rate if one transmitter sent continuously over every available distinct path, is low. These networks, such as the Internet, are optimized for high response time at the expense of lower throughput. Conversely, in data collection networks of the type, topology, and other characteristics described herein, it is both feasible and desirable to drive throughput (and utilization) as close to 100% as possible, when measuring transmissions from the edge of the network to a central data collection point.

It is well known in the art that if it is possible to predict how much data an endpoint (or edge device) will transmit, and how frequently transmissions must occur, then the most efficient type of network protocol is one called a slotted protocol. In a pure slotted protocol, each edge device is assigned a specific set of times and durations, applied on a cyclical basis, when it is permitted to transmit. These times and durations, called slots, are assigned in such a way that a transmission from on edge device never collides with a transmission from any other edge device using the same physical and frequency-based channel. In this model, the theoretical edge-to-hub network utilization achievable is 100% with no message loss.

The theoretical utilization can only be approached if the system clocks on every edge device are precisely synchronized. In the simplest case, this is accomplished by having all the edge devices able to receive a broadcast timing pulse that each device receives simultaneously. This is not always easy to achieve even in high-bandwidth two-way networks, with the result that prior-art algorithms such as the Network Time Protocol (NTP) have been invented to compensate for network latency variations in receiving timing signals. However, NTP is usually not a feasible solution in the special type of network described herein, because it requires too much two-way communication.

SUMMARY OF THE INVENTION

The present invention comprises a system of edge devices called Intelligent Communicating Devices (ICDs), which transmit using a slotted protocol according to a predefined or centrally assigned schedule, on a network typically lacking in carrier sense. These ICDs transmit to a Server at a centrally located data aggregation point with a Receiver capable of receiving transmissions from ICDs on at least one channel, but only capable of receiving a single transmission per channel at a given time. The Server also controls a Central Transmitter (CT) which can be used to send messages to the ICDs, both to assign transmission slots and other policies to ICDs (provisioning), and to provide timing information to the edge devices to ensure that transmissions by devices sharing the same communication channel do not collide. Access by the server to the CT is constrained.

Available CT technologies are of at least two types. One type, those based on true broadcast technology, are capable of broadcasting to all ICDs simultaneously, and may initiate a broadcast at any time. On-grid transmitters fall into this category, but so do some radio-frequency technologies. A second type of CT technology is based on conventional two-way network architectures and uses conventional two-way protocols such as TCP/IP. CTs of the second type are typically not capable of broadcasting to all ICDs simultaneously, and are not capable of initiating contact with any ICD. CTs of the second type are limited to responding to contacts initiated by an ICD. In both architectures constraints exist that limit the timing and frequency of messages from the CT to the ICDs. Methods are disclosed whereby the ICDs may be synchronized despite the constraints imposed by the characteristics of the CT type in use.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates a potential organization of a channel on which transmissions by ICDs are governed by a slotted protocol, and the channel is used by multiple synchronized endpoints.

FIG. 2 illustrates an on-grid network where the CT broadcasts such that all ICDs receive every broadcast simultaneously, but where the number of broadcasts is constrained and the specific time of a broadcast cannot be guaranteed.

FIG. 3 illustrates a network where the transmission medium used by the ICDs is different than the data-bearing transmission medium used by the CT, where the CT's data-bearing medium is a conventional network with variable latency, and where the CT cannot initiate a transmission to an ICD.

FIG. 4 illustrates the software model for the transmission queue of FIG. 2 with a true broadcast CT.

FIG. 5 illustrates the software model for the transmission queue of FIG. 3 with simulated broadcast.

DETAILED DESCRIPTION OF THE INVENTION

The present invention is directed generally toward the field of network management and provisioning, and comprises a system of edge devices called Intelligent Communicating Devices (ICDs), which transmit using a slotted protocol according to a predefined or centrally assigned schedule, on a network typically lacking in carrier sense. These ICDs transmit to a Server at a centrally located data aggregation point with a Receiver capable of receiving transmissions from ICDs on at least one channel, but only capable of receiving a single transmission per channel at a given time. The Server also controls a central transmitter (CT) which can be used both to assign transmission slots to ICDs, and to provide timing information to the edge devices to ensure that transmissions by devices sharing the same communication channel. Access by the server to the CT is constrained. Among the reasons why access is constrained are that the cost of the CT device and associated infrastructure is high enough that the system is effective only if shared by multiple applications and that the bandwidth of the CT, or the share of the bandwidth of the CT available to the system of the present invention, is limited. In some CT architectures both these constraints and others may apply. Typically the CT cannot guarantee the exact transmission time of any message. Methods are disclosed whereby the ICDs may be synchronized despite the constraints imposed by the characteristics of the CT. Distinct methods are required depending on the specific capabilities of the CT device and infrastructure which are in place. If the CT is capable of true broadcast, initiating the transmission of a message to all CTs simultaneously at a time unspecified and unknown to the ICDs, then one method is used. If the CT cannot initiate such a broadcast but allows two-way communication, then another method is used.

The distinct workings of the CT system may be hidden from the application software on the Server and on the ICDs from differences in messaging behavior resulting from the capabilities of the CT.

Endpoint Behaviors

FIG. 1 illustrates a segment in time of the utilization of an edge-to-aggregation-point data-bearing channel 1, which is organized according to a method of the present invention, and which is presently utilized by a collection of five ICDs. The time line of the channel 1 is organized into an indefinitely recurring sequence of Master Frames which are further subdivided into Slots 3. The segment of time in FIG. 1 comprises an entire Master Frame 2 beginning at Master Frame Origin 5 and ending with the following Master Frame Origin 6. Each occupied slot in FIG. 1 is marked with the ID of the ICD which is granted a Permission 4 to transmit in that slot. A Permission 4 is expressed in terms of the offset in slots for the first transmission with each Master Frame, the Interval 7 between transmissions, and the maximum number of transmissions permitted within a Master Frame. A Permission remains in force until it is revoked, so that an ICD that has sent its maximum number of transmissions in the current master frame may begin transmitting again after the next Master Frame Origin (MFO). There is no requirement that a next Master Frame originate immediately following the last slot in a first Master Frame. Dead or unused time in the Master Frame duration is permitted. This allows for application-specific adjustments in the Master Frame Origin, such as keeping synchronized with external environmental factors such as the diurnal cycle. The Origin in time of each Master Frame is announced by the CT in advance of the time when the Master Frame actually begins.

Although ICDs are granted permission to transmit in a designated slot or collection of slots within each master frame in perpetuity or until the grant is revoked, an ICD does not infer the origin of the next Master Frame based solely on its time clock. An ICD will only transmit when all the following conditions are met: a) it has received a Master Frame Announcement (MFA) from the CT indicating when the next Master Frame will begin; and b) according to the ICD's local time clock the announced Master Frame has begun; and c) according to the ICD's local time clock a slot has arrived in which the ICD has permission to transmit.

The Master Frame Announcement, which is generated by the Data Collection Server, alerts endpoints in advance as to the future time of the origin of the next upcoming Master Frame. In its simplest form, the Master Frame Announcement needs to contain only three data elements: the Packet Type, the precise future time of the next Master Frame origin, and the precise time at which the transmission of this Master Frame Announcement began. The MFA may contain other information dictated by the network topology or network policy. The MFA that contains the time of the next Master Frame Origin (MFO) may be transmitted as soon after the origin of the current Master Frame as the time of the next MFO is known. Further, the MFA may be broadcast multiple times during one Master Frame, reducing to near zero the probability that some otherwise functional endpoint will fail to receive the MFA.

The implementation of a Broadcast Transmission Queue may depend on the capabilities of the CT, but the interface presented to the applications or other applications that place messages into the Broadcast Transmission Queue is uniform regardless of the type of CT in use. While the general concept of a transmission queue is well known in the art, achieving endpoint synchronization regardless of the type of CT in use requires appropriate management of the transmission queue.

Embodiment 1. In a first embodiment of the invention, FIG. 2 illustrates an electrical distribution substation 21 from which originates an electrical distribution grid 26. The substation 21 contains a Server 23, a CT 24, and a Receiver 25. The Server has conventional two-way network communication with the CT and Receiver via Local Area Network 27. The network medium for the slotted edge to aggregation point transmissions used by the ICDs is the electrical distribution grid 26, and the network medium for transmission from the aggregation point to the edge is also the electrical distribution grid 26. At or near the edge of the electrical distribution grid are attached one or more ICDs 22. The one or more ICDs contain a system clock, a processor, storage for programs and data, and a transceiver capable of sending messages on the electrical distribution grid 26 to Receiver 25, and receiving messages sent by CT 24. The network medium provides at least one Edge-to-Aggregation point (E2A) channels, and one Aggregation-to-Edge-Point (A2E) channel. In this embodiment, the A2E channel is a broadcast channel and messages from the CT 24 are received by all ICDs attached to the electrical distribution grid 26 simultaneously with respect to the tolerances of the slotted protocol used to arbitrate the use of the multiplicity of slotted E2A channels among the ICDs on the electrical distribution grid.

A collision is defined as a transmission from a first ICD coinciding with or overlapping with the transmission of a second ICD, such that the message transmitted from neither ICD is decipherable at the Receiver 25. To prevent collisions, all the ICDs 22 which share a common E2A network medium (here, the electrical distribution grid 26 originating at substation 21) must be synchronized in time to within the tolerances of the slotted protocol.

Some of the prior art on collision prevention in similar contexts focuses on having the endpoints (here, ICDs) engage in collision avoidance by sensing the transmissions of other endpoints. However, because of the physical nature of the electrical grid as a transmission medium, these methods are not generally effective. Most ICDs cannot reliably detect the activity of most other ICDs. In some cases an ICD will be unable to detect the activity of any other ICD.

Other prior art relies on a shared broadcast system clock whose “ticks” (precise synchronous announcements of the time or markers of time intervals) are available to all endpoints. The prior art requires a dedicated and relatively high-bandwidth broadcast channel, and is the solution used (for example) in some wireless RF networks. However, in this embodiment of the invention, the CT 24 can only transmit at low frequencies, and a multiplicity of applications are placing messages in the CT's job queue so that it cannot be considered to be dedicated to the task of synchronization. Consequently, if the prior art method of synchronizing the ICDs by transmitting a timing pulse at the exact Master Frame Origin were used, then it would often become necessary to interrupt other transmissions in progress to do so. Because the A2E channel must be managed as a scarce resource, this is undesirable.

According to this embodiment, the A2E channel is constrained to have the following characteristics not typical in the prior art: The number of broadcasts permitted per hour is highly limited, in some cases to 10 per hour or even fewer. The limitation may be imposed due to duty cycle of the transmitter or due to the cost of transmitting.

The length of a broadcast burst is sharply constrained, often to as little as a few hundred bits. The precise timing of a CT transmission cannot be guaranteed because the A2E channel and CT are shared by multiple applications. Timing transmissions such as Master Frame Announcements can be prioritized to the head of the transmission queue, but cannot interrupt a transmission in progress. On-grid transmitters operating under the constraints described in this embodiment are well known in the art. An Audio Frequency Ripple Controller (AFRC) is an example of such a device.

FIG. 4 illustrates an implementation of the Broadcast Transmission Queue for a CT constrained as described in this Embodiment of the present invention. The implementation consists of a Queue Server 45 that accepts timing messages from a Framing Agent 46 and also from other applications 47. Queue Server 45 places messages into a storage area called the Queue 43 according to rules derived from the Service-Level Agreement (SLA) or agreements between the Queue Server and the Framing Agent 46 and other applications 47. These service levels specify a Contractual Maximum Wait Time 49, which may be different for every application or message type. FIG. 4 illustrates how the Queue Server 45 places the Master Frame Announcement 48 into the Queue 42 ahead of other messages already waiting in order to ensure that the MFA will be transmitted within the Contractual Maximum Wait Time 49 allowed by the Queue Server's contract with the Framing Agent. Note that messages accepted by the Queue Server are additionally subjected to a maximum message length, which is also part of the SLA between the Queue Server and the Framing Agent or application. The Queue Server will reject messages which are too long, or which cannot be transmitted without violating the SLA.

Whenever the CT 43, subject to its constraints based on duty cycle or frequency of transmission or cost, is ready to begin transmission, the Broadcast Server 41 examines the end of the Queue 43. If a message is waiting, the Broadcast Server removes the message from the Queue, appends to it the current time, and passes it to the CT 43. The CT immediately begins transmitting the message by modulating the A2E Broadcast Medium 44.

The organization of the Message Queue is a modified FIFO (first-in-first-out). Most messages that enter the queue have no guaranteed delivery time and no Contractual Maximum Wait Time. Some message types have associated Contractual Wait Times and Priorities. Only the Contractual Wait Time of a Master Frame Announcement is actually guaranteed. When a message with a SLA arrives at the Message Queue, the Queue Server places the message as far back in the queue as possible, such that:

• • 1. Its Contractual Wait Time is not violated, and
  • 2. It is not placed so that the Contractual Wait Time of another waiting message of equal or higher priority is violated.

Each time a priority message is placed in the Message Queue, Queue Server must scan the entire Queue to determine if the placement needed to meet the message's Contractual Wait Time will cause the Contractual Wait Time of other messages in the queue to be violated. If it is determined that a violation will occur, the disposition of the new message is determined by a rule provided by the requestor (sender): the message can be discarded with a notification to the requestor, or the message can be placed in the best available position in the queue even though that position does not meet the Contract. In the latter case, the requestor is notified of the actual expected transmission time.

If the placement of a priority message in the queue will cause the violation of the contract on older messages of lower priority, the disposition rule associated with the affected messages can be reapplied. This may cause some messages to be removed from the queue, and will cause additional notifications to be sent to the affected requestors.

Normally, a message in the process of being transmitted is never displaced by the arrival of a new message, regardless of priority.

Master Frame Announcements are the highest priority message, and they are the only messages at that priority. Thus, under ordinary circumstances, a MFA cannot be displaced, and its Contractual Wait Time is an absolute guarantee as long as said Contractual Wait Time is longer than the time required for transmitting the longest allowed message, given only that the CT serves only a single source of MFAs.

Referring again to FIG. 2, the CT 24 can be sited at the electrical distribution substation 21, or at a remote broadcast point (not shown) serving multiple substations. In the former case, the Queue, the Queue Server, and the Broadcast Server all reside on the Server 23 at the substation. In the latter case, however, where the CT at a remote broadcast point serves multiple substations, a computer called the CT Server must be dedicated to the CT and host the Queue, a Remote Queue Server, and the Broadcast Server. The substation Server 23 hosts only a local Queue Server, which is in two-way high-speed communication via a conventional wide-area network or the Internet with the Remote Queue Server. The Local Queue Server offers the Broadcast Queue interface to the Framing Agent and the other applications, which are not required to behave differently because the CT is a shared resource.

When the CT is a shared resource, there are two possible arrangements regarding framing. In the first solution, all the Data Collection Servers share the same Master Frame Cycle and Slot Size. One Data Collection Server is designated as the sole generator of the Master Frame Announcements, and only that Server hosts a Framing Agent.

In the second solution, the Data Collection Servers can have different Slot Sizes and Master Frame Cycles, but the Master Frame Sizes and Master Frame Origins are selected so that two or more Master Frame Announcements never arrive at the CT within one predefined interval, where said interval is equal to or greater than the Contractual Maximum Wait Time of the MFA for the Server with the shortest Contractual Maximum Wait Time. In this case, each Server at a Data Aggregation Point hosts its own Framing Agent. The identity of the Data Collection Server must be included in the MFA, to ensure that ICDs do not obey directives issued from a Server other than the one properly associated with said ICDs.

When a Master Frame Announcement reaches the front of the broadcast transmission queue and the previous message has been completely transmitted, then the starting transmission time of the MFA is exactly known by the Broadcast Server. At that point the Broadcast Server inserts the transmission start time into the MFA after the data provided by the originating Server. Based on the assumption that the transmission latency is the same at all receiving endpoints, the receiving endpoints can use the transmission start time in the message to eliminate any drift from their local clocks, and then reset their transmission schedules based on the upcoming future beginning of the new Master Frame. This ensures that the endpoints are synchronized to a tolerance based on the variation in latency introduced by the transmission medium employed by the broadcast CT.

This tolerance can be calculated or measured for a given broadcast transmission medium. Said tolerance determines the slot size with respect to the length of message packets transmitted in the slots by the ICDs. The closer the tolerance achieved, the higher the utilization achieved on the slotted channel.

Embodiment 2.

In a second embodiment of the invention, no in-band broadcast channel is available, and a third-party network provider is substituted to simulate the in-band CT and receivers at the ICD. This embodiment is called a simulated broadcast embodiment. FIG. 3 illustrates an electrical distribution substation 31 from which originates an electrical distribution grid 36. The substation 31 contains a Server 33, a wireless transceiver acting as the CT 34, and a Receiver 35. The Server has conventional two-way network communication with the CT and Receiver via Local Area Network 37. The network medium for the slotted edge to aggregation point transmissions used by the ICDs is the electrical distribution grid 36, and the network medium for transmission from the aggregation point to the edge is any conventional network that supports TCP/IP or other two-way communications protocols, but typically is a wireless network such as the cellular cloud, represented in FIG. 3 by the presence of antennae on the ICDs 32 and CT 34. At or near the edge of the electrical distribution grid are attached at least one ICDs 32 enabled for wireless machine-to-machine communication. The ICDs contain a system clock, a processor, storage for programs and data, a transmitter capable of sending messages on the electrical distribution grid 36 to Receiver 35, and transceiver or modem capable of polling the Server 33 by means of CT 34 and receiving responses sent via CT 34. The network medium provides at least one edge to aggregation point (E2A) channel. A Simulated Broadcast Server program executing on the Server 33 may simulate the A2E channel in this embodiment. In this embodiment, other software stored and executing on the ICDs 22 poll the Server 33 on a regular, but infrequent, basis consistent with the constraints imposed by the wireless provider on machine-to-machine communications, which are well known in the art. The constraints center on the concept that the number of simultaneous polls from ICDs in a single zone of the wireless network must be a small number based on the capacity and traffic in the zone. One method of ensuring that the constraints are met involves assigning polling times to ICDs based on the geospatial location of the ICD. The ICDs are assigned designated A2E polling times with respect to the Master Frame Origin as well as E2A transmission slots when they are provisioned. All ICDs that share a Server must poll at the same frequency which is well-known to the Server.

In this second embodiment, the actual settings of the system clocks on the ICDs is accomplished using a capability of the wireless network or other out-of-band means unrelated to the A2E messaging protocol, because the latency associated with wireless and other low-cost networks using routed protocols such as TCP/IP is insufficiently consistent to be used for this purpose. Even TCP/IP application protocols such as NTP that are well known in the art and designed for the specific purpose of node synchronization have failed to achieve the desired synchronization tolerance required for the present invention. Specifically, the network providing the simulated broadcast medium must provide some mechanism for synchronization.

Some embodiments of the cellular communications network use a time-division multiplexing scheme similar to that employed by the E2A channels in the present invention, which also requires that cellular transmitters on endpoints (e.g. mobile phones) be time synchronized. This time synchronization is adequate to achieve the tolerances of the present invention if the design of the cellular receiver in the ICD does not introduce too much additional latency in exposing the network time to the software executing on the processor of the ICD.

An alternative is that a broadcast timing pulse may be employed. Several public and private RF architectures include the capability to broadcast a timing pulse detectable by a receiver in the ICD. When a timing pulse is employed, the Framing Agent pre-announces the time at which the next Master Frame begins, as before. A timing pulse is then broadcast upon the actual Master Frame Origin. In the case where the Server does not have the option to determine the time at which the pulse is broadcast, then the Framing Agent must pre-announce the actual system time of a future timing pulse, and a window of local clock times within which ICDs may expect the pulse to be detectable. The window is required to ensure that ICDs do not ignore a timing pulse or falsely synchronize on the wrong timing pulse because the local clock has drifted. Ideally, timing pulses will occur less frequently than the interval of time in which all ICDs are expected to have polled the Server exactly once.

The second embodiment requires a different method of managing the Broadcast Transmission Queue as compared to the first embodiment. Prioritizing the messages in the queue is not effective, because no ICD will receive any message until it polls, regardless of its priority. Nor can a message be removed from the queue when it is transmitted to an ICD as a polling response, because potentially every ICD must obtain the message in order to simulate a broadcast CT.

As messages are placed in the Simulated Broadcast Transmission Queue, they are assigned a monotonically ascending sequence number. This sequence number semantically replaces the timestamp inserted in broadcast messages by the True Broadcast Server. A sequence number is similarly stored in a database inventory of all ICDs served by this CT, which represents the last sequence number assigned to any message the last time each ICD polled the Server for messages. When an ICD polls the Simulated Broadcast Server, the server responds by sending to the endpoint ALL messages in the Simulated Broadcast Queue whose sequence number is higher than the sequence number currently associated with the polling endpoint, then updates the sequence number in the inventory of endpoints entry for that ICD with the highest sequence number of any message in the newest response. The sequence number need not be transmitted to the endpoints, so it can be as large as needed without adding to the bandwidth burden of the constrained Simulated Broadcast Channel. The TCP/IP protocol will ensure that the messages arrive at the endpoint, though it will not ensure that it does so with a predictable latency. This scheme ensures that each endpoint receives each message only once.

In the Simulated Broadcast embodiment, inserting the central system time in the Master Frame Announcement is not useful because the latency between the Simulated Broadcast Server and the endpoint is not constant or predictable. Instead, Master Frame Announcements must be enqueued at least 2 T prior to the time of the master frame origin, where T is the polling frequency of the endpoints.

Emergency Messages and Preemption.

In any embodiment of this invention, Master Frame Announcements are normally the highest-priority messages in the Broadcast Transmission Queue. However, in the case where the network substrate is the power grid, there exist rare events that endanger the power grid, such as an out-of-tolerance peak load or an equipment failure that is cascading. In such circumstances, extraordinary measures must be taken to protect the grid, even if the consequence is that endpoint synchronization is temporarily lost or the contract with the A2E network provider is violated. In these rare cases, the Broadcast Server is permitted to abort the message currently in transmission, if any, and to pre-empt even a Master Frame Announcement.

The Broadcast server minimizes the disruptive effect of emergency broadcasts by means of the algorithm used to re-normalize the queue after a pre-emption, as described below. It is unlikely that the preemption algorithm will be needed for radio-based solutions, but when the broadcast transmitter is a low bandwidth device such as an Audio Frequency Ripple Controller, then the preemption capability becomes necessary. When a Simulated Broadcast Server is employed, pre-emption may be simulated by sending SMS messages to the ICDs affected by the emergency conditions. Upon receipt of such a message, an ICD polls the Server immediately regardless of its place in the polling sequence.

FIG. 5 illustrates the Simulated Broadcast Queue. As in the previous embodiment, Framing Agent 56 places MFAs and potentially other messages, such as timing pulse identifications, into the Broadcast Queue 52 by means of Queue server 55. In this embodiment, Broadcast Queue 52 observes a simple FIFO discipline, except that messages are not removed from the queue until all ICDs in the Inventory of Endpoints 53 have a higher sequence number than the message. ICDs 54 poll Simulated Broadcast Server 51 at regular frequencies but staggered intervals. When Simulated Broadcast Server 51 detects a poll from an ICD, it looks up the sequence number last associated with that ICD, and responds by sending said ICD all messages in the Broadcast Queue 52 which have a higher sequence number than that associated with the polling ICD. The sequence number of the highest-sequenced message in the response is then associated with that ICD in the Inventory of Endpoints 53. The Simulated Broadcast Server then queries the Inventory of Endpoints to find the lowest sequence number associated with any ICD, and remove any messages from the Broadcast Queue whose sequence number is lower than that. It will be obvious to one ordinarily skilled in the art that the failure of some ICDs to poll on schedule could cause messages to be retained in the queue for a long period of time. It is therefore advisable to add to this method a method for detecting failed endpoints and excluding their sequence numbers from the query of the Inventory of Endpoints 53 used to determine the threshold sequence number for purging the Broadcast Queue.

Endpoint Clock Adjustments.

The ICDs employ stored policies and the information in the Master Frame Announcements to determine how ICDs utilize the multiplicity of channels so that each ICD will avoid transmitting when the transmission will collide with the transmission of another ICD.

Each ICD can transmit on a multiplicity of frequency-based channels on the power grid medium, but each ICD can only transmit on one frequency-based channel at a time. Further, each ICD transmitter has a duty cycle which requires it to be quiescent on all channels at regular intervals. The broadcast channel (received by endpoints) is both full-duplex and asynchronous with respect to the slotted channel, regardless of whether it is simulated or true broadcast.

For every slotted channel on which a particular ICD is permitted to transmit, the ICD stores a policy that determines in which slots within the Master Frame it is expected to transmit. No ICD has a need to know the slot assignments of any other ICD. The Server is responsible for provisioning the ICDs in such a way that (when properly synchronized) transmissions from no two ICDs ever collide on a slotted channel.

An ICD joining the network does not transmit at all until it has received a Master Frame Announcement. (This is another reason for transmitting the Master Frame Announcement more than once during a given Master Frame.) When an ICD receives an MFA, it performs two activities.

First, the ICD records the Master Frame Origin for the next cycle. The ICD does not distinguish between new and redundant MFAs. Any MFA received during the current Master Frame always is construed to pertain to the next Master Frame.

Second, an ICD which is the client of a true broadcast CT compares the transmit time in the MFA with the local clock time when the message was received to determine if there has been clock drift since the last MFA. If so, the endpoint computes the size and sign of the adjustment needed, and waits for the next quiescent transmitter interval. At that time, a smooth adjust technique (for example, as implemented by the Linux command “date—a”) is used to rectify clock synchronization. When such an adjustment is made during the quiescent interval, the clock adjustment cannot interfere with the correct operation of the digital signal processing firmware, and the adjustment is complete in time to ensure that the next transmission is synchronized with the global slotting scheme.

An ICD that is the client of a simulated broadcast CT does not need to compute clock drift because it is able to use the A2E network time or pulse to keep its clock synchronized.

An ICD that has not received any Master Frame Announcement for the next Master Frame by the end of the current master frame stops transmitting until it receives another MFA. This prevents the unsynchronized ICD from colliding with the transmissions of other endpoints. Because of redundant MFA transmissions, it is unlikely that any ICD would fail to receive all MFAs in an entire frame unless its receiver has malfunctioned.

Other Embodiments.

This description of the preferred embodiments of the invention is for illustration as a reference model and is not exhaustive or limited to the disclosed forms, many modifications and variations being apparent to one of ordinary skill in the art.

Claims

1. A system for synchronizing an intelligent communicating device (ICD) on an electrical distribution grid, comprising: a Central Transmitter (CT) configured to be coupled to the electrical distribution grid on a high voltage side of a transformer; andan ICD comprising a clock and configured to be coupled to the electrical distribution grid on a low voltage side of the transformer;wherein: the CT is configured to transmit a plurality of master frame announcements (MFAs) at arbitrary times through the transformer to the ICD, each MFA identifying a time of transmission of the respective MFA and a future time of a corresponding master frame, the corresponding master frame comprising a plurality of time slots, each time slot associated with a particular ICD of a plurality of ICDs; andthe ICD is configured to: receive one MFA of the plurality of MFAs;based on the time of transmission of the one MFA and the future time of the corresponding master frame, synchronize the clock;await the future time of the corresponding master frame; andtransmit data on the electrical distribution grid to the high voltage side of the transformer in a time slot of the corresponding master frame that is associated with the ICD.

2. The system of Claim 1, where the CT is an on-grid transmitter.

3. The system of claim 1, where the CT is a private radio transmitter.

4. The system of claim 1, where the CT is simulated by a wireless two-way network.

5. The system of claim 1, wherein the ICD comprises a receiver configured to communicate with a type of the CT.

6. The system of claim 1, wherein the CT is controlled by a software system comprising a Broadcast Queue Server, a Broadcast Transmission Queue and a Broadcast Server.

7. The system of claim 1, wherein the electrical distribution grid lacks a carrier sense.

8. The system of claim 1, wherein the plurality of MFAs are transmitted on a broadcast channel, and the ICD transmits the data upstream through the transformer on an edge-to-aggregation-point channel that is a different channel than the broadcast channel.

9. A system for synchronizing an intelligent communicating device (ICD) on an electrical distribution grid, comprising: a server comprising a processor, the server configured to: maintain, on a high voltage side of a transformer, a queue of messages, at least one of the messages comprising a master frame announcement (MFA) that identifies a future time of a corresponding master frame, the corresponding master frame comprising a plurality of time slots, each time slot associated with a particular ICD of a plurality of ICDs located on a low voltage side of the transformer; andan ICD configured to: request, via a communication medium other than the electrical distribution grid, the messages from the queue of messages;receive the MFA:receive synchronization information to synchronize a clock of the ICD;await the future time of the corresponding master frame; andtransmit data on the electrical distribution grid to the high voltage side of the transformer in a time slot of the corresponding master frame that is associated with the ICD.

10. The system of claim 9 wherein to receive the synchronization information to synchronize the clock of the ICD, the ICD is further configured to request, from a synchronous central system time, a current time.

11. The system of claim 9 wherein to receive the synchronization information to synchronize the clock of the ICD, the ICD is further configured to: receive in the MFA a time identifier that identifies a future time of a timing pulse;receive the timing pulse; andset the clock of the ICD to the future time in response to receiving the timing pulse.

12. A method for synchronizing an intelligent communicating device (ICD) on an electrical distribution grid, comprising: transmitting, by a Central Transmitter (CT) coupled to the electrical distribution grid on a high voltage side of a transformer, a plurality of master frame announcements (MFAs) at arbitrary times through the transformer to a plurality of ICDs that receive power from the transformer, each respective MFA identifying a time of transmission of the respective MFA and a future time of a corresponding master frame, the corresponding master frame comprising a plurality of time slots, each time slot associated with a particular ICD of the plurality of ICDS;receiving, by a first ICD of the plurality of ICDs, one MFA of the plurality of MFAs, the first ICD comprising a clock and coupled to the electrical distribution grid on a low voltage side of the transformer;based on the time of transmission of the one MFA and the future time of the corresponding master frame, synchronizing the clock;awaiting the future time of the corresponding master frame; andtransmitting, by the first ICD, data on the electrical distribution grid to the high voltage side of the transformer in a time slot of the corresponding master frame that is associated with the first ICD.

13. The method of claim 12, further comprising receiving, at a server coupled to the CT, a plurality of messages from a plurality of different applications, wherein some of the plurality of messages comprise respective MFAs of the plurality of MFAs.

14. The method of claim 13, further comprising reprioritizing the plurality of messages to prioritize the plurality of MFAs based on a Contractual Maximum Wait Time.

15. The method of claim 13, further comprising notifying an application of the plurality of different applications that a message of the plurality of messages cannot be broadcast before a Contractual Maximum Wait Time has elapsed.

16. The method of claim 12, wherein the first ICD will cease to transmit if, at the end of a Master Frame, the first ICD has not received a Master Frame Announcement announcing a time of a next Master Frame.

17. A method for synchronizing an intelligent communicating device (ICD) on an electrical distribution grid, comprising: maintaining, on a server on a high voltage side of a transformer, a queue of messages, at least one of the messages comprising a Master Frame Announcement (MFA) that identifies a future time of a corresponding master frame, the corresponding master frame comprising a plurality of time slots, each time slot associated with a particular ICD of a plurality of ICDs located on a low voltage side of the transformer;requesting, by an ICD of the plurality of ICDs via a communication medium other than the electrical distribution grid, the messages from the queue of messages;receiving, by the ICD, the MFA;receiving, by the ICD, synchronization information to synchronize a clock of the ICD;awaiting the future time of the corresponding master frame; andtransmitting, by the ICD, data on the electrical distribution grid to the high voltage side of the transformer in a time slot of the corresponding master frame that is associated with the ICD.

18. The method of claim 17, further comprising assigning each message of the plurality of messages a monotonically increasing message sequence number unique to the queue.

19. The method of claim 18, further comprising maintaining an inventory of the plurality of ICDs.

20. The method of claim 19, further comprising associating with an inventory record of each ICD a highest message sequence number which each ICD has already received.

21. The method of claim 20, wherein requesting, by the ICD of the plurality of ICDs via the communication medium other than the electrical distribution grid, the messages from the queue of messages further comprises periodically requesting, by the ICD of the plurality of ICDs via the communication medium other than the electrical distribution grid, the messages from the queue of messages.

22. The method of claim 21, further comprising providing, in response to a request from the ICD, messages in the queue which have a higher message sequence number than a highest message sequence number associated with the ICD.

23. The method of claim 22, further comprising updating, the inventory record of said ICD with the highest message sequence number transmitted to said ICD.

24. The method of claim 23, further comprising removing from the queue each message which has a lower message sequence number than a highest message sequence number of each ICD of the plurality of ICDs.

25. The method of claim 17, further comprising removing from the queue messages which have resided in the queue for longer than a predefined number of polling cycles, where a polling cycle is an interval of time in which all ICDs of the plurality of ICDs are expected to have polled once.

26. The method of claim 17, wherein receiving, by the ICD, the synchronization information to synchronize the clock of the ICD comprises requesting, from a synchronous central system time, a current time.

27. The method of claim 17, wherein an alternative protocol is used to synchronize a plurality of clocks of the plurality of ICDs.

28. The method of claim 17, wherein receiving, by the ICD, synchronization information to synchronize the clock of the ICD comprises: receiving in the MFA a time identifier that identifies a future time of a timing pulse;receiving the timing pulse; andsetting the clock of the ICD to the future time in response to receiving the timing pulse.

29. The method of claim 17, wherein each ICD initiates a polling cycle of the each ICD with a different offset from the master frame than any other ICD of the plurality of ICDs.