WO2004109548A1

WO2004109548A1 - Improvement in relation to storing and transmitting a positional message

Info

Publication number: WO2004109548A1
Application number: PCT/FI2004/050084
Authority: WO
Inventors: Markus Mikkolainen; Roope Anttinen
Original assignee: Indagon Oy
Priority date: 2003-06-09
Filing date: 2004-06-08
Publication date: 2004-12-16
Also published as: FI20035087A0; EP1636720A1; FI20035087A; FI116026B

Abstract

A method and system for storing the positional data of a mobile station in at least two servers, such that the mobile station transmits reports of positional data alternately to various servers. In addition, the dependent claim requires that the system be self-organizing and the system segments be independently functional.

Description

Improvement in relation to storing and transmitting a positional message

The invention relates to storing in servers a positional message transmitted by a mobile station. In a typical operating environment, for example, a mobile station effects determination of its own location by using, for example, GPS positioning. Positional information is processed and stored in servers, with which a mobile station is in communication over a mobile telephone network by using, for example, a GPRS, SMS, Tetra, satellite, UMTS or some other such link for the transmission of a positional message.

Typically, positional information is needed for example in delivery traffic, haulage, ordering systems for taxis, and routing. Thus, the mobile stations integrated with a system transmit their positional information to the system, which uses the positional information and its history, for example, for dispatching orders to the nearest mobile station, for example to a delivery truck or a taxi. Typically, the system comprises a plurality of mobile stations, a smallish number of servers for storing positional information and servers to drive applications, or workstations which read positional messages from said storing servers. Applications can also be performed in the very servers that the positional information is stored in. A positional message can also be transmitted on the basis of an inquiry from a server.

Designing an extensive positional information system creates problems because of a huge number of transactions, and a high reliability normally requires duplex or doubled systems. The increasing number of transactions necessitates that the load be balanced between several servers. Generally, different users are allocated to different servers, thus reducing a load per server. These solutions do not offer a cost-effective way of providing simultaneously load balancing and a high accessibility in a malfunction incident. In addition, the inventive method enables updating of the system and a controlled recovery from a malfunction, yet without requiring a doubled backup capacity. The objective is a high accessibility, yet without wasting any computer capacity for doubling.

The inventive system and method for storing a positional message functions in such a way that a mobile station itself stores positional information in a predetermined manner in a number of servers, such that each server is stored with some of the

information in view of delivering sequential positional messages to different servers. If there are several servers, the inactivation of a single server only disables a small portion of stored information. Since a positional message consists of sequential measuring results not very different from each other, the inactivation of one server does not significantly undermine the available measuring information. Thus, availability is retained, even if one of the servers should fail or be disabled because of updating or maintenance. Thus, in practice, the inactivation of a single server means that the updating time of a positional message to be read is doubled at least in some of the measurements.

Since each mobile station writes its messages alternately for two or more servers, each server will have a capacity of serving several mobile stations. As a message is not doubled, the amount of capacity spent is not more than what is used in single non-doubled servers. The system distributes a load between several servers automatically and, furthermore, the system expansion and server replacements are possible without shutdowns.

A system of the invention will now be described with reference to the figures

Rg. 1 shows a doubled system of the prior art fig. 2 shows a system provided with load balancing of the prior art fig. 3 shows how a system of the invention operates fig. 4 shows the operation of a more extensive system

Referring to all figures, mobile stations ml and m2 transmit their positional information to servers indicated by references with a prefix s. Messages or communications containing positional information are represented by arrows.

Rg. 1 shows a doubled system, in which every server sla is provided with a server sib as a backup server. The system stores each positional message in both servers. Consequently, this requires twice the number of servers compared to a non-doubled system, having one server designated for each mobile station. Reading and writing speeds are equal in the basic mode, yet it is possible to effect the reading also from both servers, resulting in a doubled reading capacity. In practice, however, the limiting factor is the speed of writing and, thus, the system of fig. 1 does not

function any faster, even with a doubled number of servers. In the event of malfunction, the failure of one server does not affect operation and, thus, a higher functional reliability is achieved by doubling. A typical bottleneck is the number of messages to be written from several writing stations at the same time.

A system provided with load balancing is shown in fig. 2. In that, mobile stations ml and m2 transmit their positional messages a and b to a load-balancing handling device LB, which in turn dispatches the information to various servers. The positional message of a mobile station is only stored in one server at a time and, thus, if a server fails, the worst result is that the positional information of some mobile stations is lost completely. There is no loss of capacity as a result of doubling but, on the other hand, the load-balancing device LB constitutes another potentially failing component. In a malfunction incident, some of the information may be totally lost, even if just one server were disabled.

Rg. 3 illustrates a system of the invention. In that, mobile stations ml and m2 transmit independently positional messages al, a2, a3, bl, b2 and b3 to servers. The mobile units use a remainder, calculated from a time of day produced by their own ordinal, a positional message ordinal or a GPS clock, for the determination of a server, and information is provided by the servers in regard to which servers each mobile station is to send its positional information and on which basis. When using a time of day, the remainder is calculated from time slices, whose duration is matched relative to a time gap between measurements in such a way that, on the one hand, sequential measurements are likely to arrive in different servers, but most time gaps shall receive a measuring result. The simplest approach is to use a basis of division calculated solely on the basis of the ordinal of a mobile station. Hence, the application programs know, even without inquiry, on the basis of the number of a mobile station, where the positional information of each station is located.

The inventive method of placing sequential positional information in various servers makes it possible that the servers can be located, even physically, in various parts of a network, such that the positional messages travel outside a mobile network along various routes, whereby a failure anywhere between a mobile network and servers does not entail losing the entire positional information of even a single mobile station. Thus, the part of a network needed for reading a message is largely

different. Hence, in practice, the result is a fault tolerance which is better than what is achieved in a doubled system, as the system can be readily decentralized in terms of the topology of an information network and geographical location. This cannot be achieved in a doubled system, because the doubling itself is not done until after the reception of a positioning message. If storage on the basis of a time of day is used and a measuring message is transmitted at each time gap, the result is a system in which positional information can be found for application programs without experimentation and servers can be located physically in various parts of a network. Thus, the system functions as effectively as an optimally load-balanced one without extra reading times, but the positional information regarding each mobile station is distributed in a different part of a physical network. A result will be an excellent fault tolerance without compromising performance.

In several cases, the geographical decentralization may also reduce a load in an information network due to reading positional data, because the application programs are able to choose the network-topologically closest server as a server principally used thereby.

Between two servers, for example, the information or data can be stored also on the basis of a GPS time of day, the information-seeking applications being thus aware as to where the latest piece of information could be found. It should be appreciated in this case that, if mobile stations write messages, for example, over sequences or periods of 17 seconds and the writing takes place, for example, at varying intervals of about 10-30 seconds, the application seeking for information cannot be sure that the writing of a positional message would have taken place for this particular server over the latest 17-second period. However, the latest message from data distributed over a plurality of servers is thus obtained in most cases by no more than two reading times, i.e. the reading of two servers discloses the latest information provided that the positional information in the system is no more than 2*17=34 seconds old. If the first reading time provides a notably outdated positional message, i.e. no positional data has been written over the latest writing sequence, it is nevertheless likely that the latest data can be found in the preceding server.

Rg. 4 illustrates the distribution of three servers and several mobile stations amongst all servers. Naturally, the mobile stations are only capable of storing positional data, for example, in just two servers and the number of servers can be more than three. According to fig. 4, the inactivation of any server only eliminates 1/3 of positional data. In addition and if desirable, the application may request a positional message from only one server, whereby the loading is small but the positioning data is respectively more outdated and less accurate. However, this is often sufficient. For example, the order of a taxi is only worth suggesting to cars which have been in the same part of the city a few minutes earlier, the more accurate positional data for other cars is of no interest in this case. Thus, it is possible to address just one server for approximate positional information and, if necessary, to specify the positional data by requesting more updated information from other servers. If the messages are transmitted at equal intervals, it is also possible to conclude on the basis of a clock which server is in the possession of the latest positional data at any time.

Details regarding the implementation of one system of the invention in various operating modes will be described hereiπbelow. The invention is characterized by what is set forth in the claims. The inventive system may also function while configured in several different operating modes. For example, during the process of updating or inauguration, the most preferred function mode may be other than a sequential storage mode.

The system's servers may be self-configuring, thus deciding among themselves a numerical order and a configuration thereof on the basis of supplied instructions. The system's configuration can be changed without a shutdown. Thus, it is possible to start operation even with one server, to set up alongside the same a second server with doubling storage, whereby, in a moment, both will have the same copies of positional data over quite a lengthy period. Now, the earlier server can be decommissioned and replaced with another without any shutdown in services. This new server, along with the other one, can serve all mobile stations by means of sequential storage, thus providing a good compromise between storage capacity and accessibility. During a sequential storage operation, the mobile stations hence transmit every other positional message to every other server.

The inauguration and configuration of one application of the invention are based on broadcast messages transmitted by servers, on the basis of which other servers known which servers are up and, moreover, the addition or elimination of a new server can be fully automated as the servers monitor or track messages transmitted by other servers and, if necessary, modify their operation accordingly.

The system comprises a plurality of terminals, a few servers, and applications which seek data from the servers. The terminals supply the servers with positional information about themselves. The system is intended to transmit and store position as cost-effectively and reliably as possible. Some of the servers can run a position- receiving and storing program, and some a position-seeking program which communicates with an application. The system's principal communication mode is preferably UDP/IP or multicast. One way of providing a convenient and effective inaugurability and efficiency is to multicast a message about functionality at constant time intervals between servers. Carried along this message is the time of a server's activation instant. The servers' clocks are synchronized with each other at the accuracy of about a second. The servers decide among themselves their own order or sequence on the basis of the activation time. The first activated server picks up the first ordinal (ordinal 0), the last activated receives the number n-1.

Each mobile station is supplied with a message as to which server/servers to contact. The server software programs, which inquire positional servers for locations and transmit the same to applications, listen also to this transmission and thus know whom to ask for a location. If it is desirable to provide a geographical or network- topological decentralization, which is functional even when servers cannot establish communication with each other, the information about the delivery of messages or the transmission of messages failing to reach servers shall arrive along some other route, for example from mobile stations.

The inventive system will now be described in terms of its operation during inauguration and in other atypical modes. Some of the modes do not comply with the claims, yet the overall functional mode and configurability provide novel benefits over the prior art. In a non-redundant normal mode, each server deals with those mobile stations whose remainder coincides with the server's ordinal. Subsequently, the remainder of a divided with n will be given a notation a%n. Each mobile station

has been supplied with a command to make contact with a server which has the same remainder, and if such a server is not available, to make contact with a server (id+l)%n, i.e. the next server in line. Thus, in this functional mode, we know that, if we want a location from the mobile station a, it will be found in the server a%n. Hence, the system is scaled as a number of mobile stations, servers, and applications. Thus, the system has a writing capacity (storage capacity) which is n times a server's capacity, the same applying to reading capacity. The system is not fault tolerant, when one server is faulted, the information of mobile stations using that server will be lost.

If a server, in a non-redundant normal mode, happens to be decommissioned, the system proceeds to a fault tolerance mode (degraded mode) until a timeout has lapsed. During the exceptional mode, mobile stations send position normally to the servers which are in operation. The mobile stations of the inactivated server are inaccessible until they realize that the server has inactivated and make contact with a backup server. The software programs of the application side begin to send inquiries to all servers for locations upon noticing that the system is in a degraded mode. When the degraded mode is abandoned, the number of servers has changed, the ordinals have been rearranged, and the modules are being reconfigured as contact is made thereby. Even at its maximum, the system's reading capacity in a fault tolerance or degraded mode is n times the server's capacity as far as data is available, the writing capacity is n-1.

In this mode, the system is not fault tolerant and, thus, the positional information of some mobile stations will be lost completely until the system has reconfigured itself.

In this mode, a new server will be adopted to service in such a way that, as the server is introduced, the system proceeds to a degraded mode until a timeout has passed (i.e. all servers have reconfigured themselves). Mobile stations establish normal communication with servers which they have been configured to make contact with. During the course of a degraded mode, the application servers are in a worst case forced to ask all servers for a location. When the degraded mode is abandoned, the number of servers has changed and the ordinals have been rearranged. Mobile stations will be re-configured as the make contact the next time. The system has a storing capacity which n (times the server's capacity) and

increases to n+1 as the mobile stations become reconfigured. Reading capacity varies according to implementation. Redundancy in the system remains at the same level.

The system has its safety factor increased and mobile stations coupled with the system adopt the use of cyclic storage between several servers, such that the system is first momentarily in a degraded mode (if resulting positional data is older than the last degraded mode of a cluster, inquiry is made to all servers). Mobile stations are commanded to use, for example, the cyclic configuration of 2 servers as communication is re-established thereby. As mobile stations proceed to the use of cyclic storage, the system progresses to a mode of n/A clusters (in this case, π/2).

Benefits gained by the above-described cyclic storage include:

- If a system is augmented with a new server, this is managed by a brief automatic visit to a degraded mode.

- If a system is to eliminate a server, the system visits briefly in a degraded mode, functionality suffers just a brief period, and the system's capacity falls to a server capacity n-1, the old positional data must perhaps be found at the capacity of 1, but old positional data is needed extremely rarely. Furthermore, applications can optimize the reading of old data by concluding or requesting the old numbers and transition times of the servers.

- The capacity of all servers is available for storage at all times, while in a conventional doubled n-server system, the available capacity at its maximum would be that of n/2 servers.

An extensive system, consisting of n servers and performing a cyclic storage between A servers, functions as described above, but each mobile station is configured to send its position alternately to servers id%n and (id+l)%n...(id+(A- l))%n. This makes sure that the positions or locations are distributed evenly among the A number of various servers.

The system's storage capacity is still equal to the number n of servers, reading capacity is n/A, if the latest position is absolutely necessary. If slightly older positional information is acceptable, the reading capacity is n. Unlike the previous case, the system can afford to drop 1 server and, as a consequence, the portion 2/π

of mobile stations loses its positional data at every A. Every mobile station is still accessible and can be supplied with positional data, which at its maximum is 2 transmission intervals old.

The following description deals with the operation in various situations regarding a system of the invention performing cyclic storage between the A servers of n servers.

If one server is inactivated from service, the system proceeds to a degraded mode. Ordinals are rearranged. For the duration of a degraded mode, the system has a reading capacity which at its maximum is (n-l)/A. Writing capacity is n-1. Redundancy in the system is momentarily poorer, but after the rearrangement of ordinals the positional data is again copied for two different servers, while old positional data is not necessarily available in more than one server.

Since mobile stations are in possession of A number of addresses for various servers, the mobile stations shall very quickly find an active server and receive a new functional configuration therefrom.

When a server is introduced in a system, the system proceeds momentarily to a degraded mode. Ordinals will be rearranged. Reading capacity is no more than n and writing capacity is n, and progresses towards n+1 as the system stabilizes. Redundancy in the system is the same as before. Reading capacity for old positional data suffers at least briefly. Mobile stations are reconfigured as contact is made thereby.

Safety factor for servers is lowered:

The system proceeds to a degraded mode for just a moment, mobile stations are reconfigured at the time of next communication. Old positional data remains where it was, the new data being only sent to A-1 servers. The retrieval capacity of new positions is n, the old ones have a lower capacity. The system stabilizes relatively quickly.

Safety factor for servers is heightened: The system proceeds to function the way of a system with a higher safety factor. Mobile stations are reconfigured at the time of next communication. Position retrieval functions from the beginning at the efficiency/inefficiency of a new system.

An advantage gained by the described system is that the system's storage capacity is n, while the storage capacity of a system which is doubling or copying in multiple servers would only be n/A. The system's reading capacity is nevertheless n/A, and for less accurate positional data it is n. One server may break down from the system and still a sufficiently good position is available for all mobile stations. In addition, the system knows how to reconfigure itself after the breakdown of a server.

If compared to a load-balanced system, the described system is considerably better, since in a load-balanced system, a portion 1/n of all mobile stations would be totally lost upon the failure of one server. It is easy to introduce or eliminate servers to or from the system, and the system's safety factor can be adjusted as necessary without a shutdown in service.

A cluster of N servers, wherein a backup factor B functions in such a way that every mobile station transmits its positional data to a server specified for each station, from which it is copied by the B-l servers for the next server. Hence, this is a known doubled system, a server of the invention being possibly familiar with this as well, since the system is designed to be inherently configuring and flexible during operation. If the identification of a mobile station is id and the backup factor is B, a copy of all data will always be sent to servers id%n...(id+B-l)%n. The system behaves as normal for a system secured with a B factor. The system's storage capacity is n/B, the system's reading capacity is n. If the system drops one server, nothing of the data will be lost, at least B+l servers must drop before any data begins to disappear.

When a copying system is provided with a new server, the system proceeds to a degraded mode, ordinals are rearranged, mobile stations are reconfigured. As the reconfiguration of mobile stations progresses, the system is turning into an (n+l)/B system.

When a server Is eliminated from a system functioning like this, no data will be lost, the system proceeds to a degraded mode, ordinals are rearranged, mobile stations are configured. As the configuration of mobile stations progresses, the system is turning into an (n-l)/B system. Old data is not generally re-secured, new data is secured B-fold.

When the safety factor of a copying system is heightened, the system begins to proceed to an n/(B+l) mode. Old data is not generally multiplied, new data turns into (B+l) between servers, respectively as the safety factor of servers is lowered, the system begins to proceed to an n/(B-l) mode, as described above. Old data is not touched, new data is secured (B-l)-fold.

An advantage gained by the copying functional mode is that the system operates the same way as a conventional double (B-timed) cluster, yet the cluster configuration is relatively easy to modify during operation. All parties are still aware of the location of positional data without special inquiries. If one server breaks down in the system, it does not function like conventional doubling (failing to re-double) but, instead, drops capacity and doubles again. Reading at its optimum is naturally B rimes the capacity of one server. If various mobile stations use all servers as primary servers, the momentary writing capacity can be higher than n/B. However, copying depletes capacity in any case, and so the capacity lies somewhere between n and n/B. Introducing a new server into the system keeps the doubling factor unchanged, but increases the number of servers.

The above-described system functions by using multicast in a communication between servers. This simplifies inauguration of the system, but it is also possible to supply a new server with the address for at least one existing server, which the new server can approach for necessary information. If multicast is not used, the servers must exchange configuration data with each other and check each other's functional mode from time to time with some other method. If the servers are geographically decentralized and communication links therebetween can be monitored from outside, it is necessary to employ a high-grade encryption for protecting communication. Generally, the equipment is linked to a circularly secured, dedicated local network, which only application program servers are able to ask for positional data, other traffic is blocked.

The system is characterized by an excellent compromise between fault tolerance and performance regarding the storage of positional information. The system is self- organizing, by virtue of which a faulted system functions always as well as possible and the operation improves automatically as a new server is introduced into the system. All that mobile stations need to known about the system is the addresses for sending positional data thereto. With the possible exception of the time spent in a degraded mode, the application programs are aware of which server to approach to retrieve information for each mobile station. The network data of a mobile station is always brought up-to-date automatically upon a first post-change communication with a server.

Claims

1. A method for storing the positional data of a mobile station in at least two servers, characterized in that the mobile station transmits reports of positional data alternately to various servers.

2. A method as set forth in claim 1 for storing positional data, characterized in that the mobile station transmits the data alternately to a number of servers, said number being defined by a message sent to the mobile station.

3. A method as set forth in claim 1 for storing positional data in at least two servers, characterized in that the servers, in which the sequential messages end up, are located in various parts of an information network.

4. A method for reading from servers the positional data stored as set forth in any of the preceding claims, characterized in that one server is first used for reading less frequently updated positional data, and not until more up-todate positional data is needed, is the data retrieved from all relevant servers.

5. A method for reading from servers the positional data stored as set forth in any of the preceding claims, characterized in that the location of positional data in storage is determined by a method mutually independently repeatable in each element.

6. A system for storing positional data in at least two servers, characterized in that a mobile station is adapted to transmit, upon request, sequential reports of positional data alternately to various servers.

7. A system as set forth in claim 6 for storing positional data in at least two servers, characterized in that the servers supply each other with information about their functionality and configuration and, if necessary, modify their operation according to information received from the others, and the servers request mobile stations to send their messages according to a new configuration.

8. A system as set forth in claim 6 or 7, characterized in that servers can be added to or eliminated from the system without shutting down the system.

9. A system as set forth in any of claims 6-8, characterized in that the system segments are capable of modifying the cydicity of storage or the number of copies without shutting down the service.

10. A system as set forth in any of claims 6-9, characterized in that the system servers communicate with each other by using multicast, whereby no preliminary information is needed about other servers or system.

11. A system as set forth in any of claims 6-9, characterized in that the system servers exchange with each other information about configuration on the basis of the address for at least one previously known server.

12. A system as set forth in any of claims 6-9, characterized in that the system modifies its operation automatically to comply with a new number of servers as a server is excluded or included.