DE10239299C1 - Telecommunications network distribution method for cost optimization of radio network sub-systems - Google Patents

Telecommunications network distribution method for cost optimization of radio network sub-systems

Info

Publication number
DE10239299C1
DE10239299C1 DE2002139299 DE10239299A DE10239299C1 DE 10239299 C1 DE10239299 C1 DE 10239299C1 DE 2002139299 DE2002139299 DE 2002139299 DE 10239299 A DE10239299 A DE 10239299A DE 10239299 C1 DE10239299 C1 DE 10239299C1
Authority
DE
Germany
Prior art keywords
cost
costs
groups
edges
neighborhood graph
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
DE2002139299
Other languages
German (de)
Inventor
Ulrich Lauther
Thomas Winter
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nokia Solutions and Networks GmbH and Co KG
Original Assignee
Siemens AG
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Siemens AG filed Critical Siemens AG
Priority to DE2002139299 priority Critical patent/DE10239299C1/en
Application granted granted Critical
Publication of DE10239299C1 publication Critical patent/DE10239299C1/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W16/00Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
    • H04W16/22Traffic simulation tools or models
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W16/00Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
    • H04W16/18Network planning tools

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The network distribution method uses generation of a proximity graph defining the traffic connections to the adjacent nodes for each base station, used for providing a corresponding minimum spanning tree, subsequently divided into sub-trees each corresponding to radio network sub-systems (RNS) of base stations (NodeB) operated by a radio network controller (RNC). The total cost of the telecommunications network is calculated for the sub-tree division and the sub-tree division process is repeated until the lowest total cost is obtained.

Description

Die Erfindung betrifft ein Teilproblem eines Netzwerkplanungsprozesses, insbesondere für UMTS-Netzwerke, bei dem Basisstationen (Node B) in Gruppen (Cluster) bzw. Funknetzwerksubsysteme (radio network subsystems bzw. RNS) aufgeteilt werden.The invention relates to a partial problem of a Network planning process, especially for UMTS networks, at the base stations (Node B) in groups (clusters) or Radio network subsystems (RNS) be divided.

Bei der Planung des Access-Bereiches in UMTS-Netzwerken muss eine Aufteilung der Basisstationen in Gruppen, die durch jeweils einen Radio Network Controller bedient werden, so erfolgen, dass möglichst die Gesamtkosten des Systems minimiert werden. Dieser Planungsprozess ist jedoch diffizil, weil verschiedene Kostenbeiträge wie zum Beispiel gruppeninterne Verbindungskosten, Kosten für Netzwerksteuereinheiten (radio network Controller; RNC) Kosten für die Anbindung zu einer nächsthöheren Hierarchieebene und Kosten für Verbindungen zwischen den Clustern bei sogenannten "soft handovers" in komplexer Weise und teilweise gegenläufig zusammenspielen. Dabei sind Planungsverfahren für GSM-Netze nicht ohne weiteres einsetzbar, da es beispielsweise beim UMTS-Zugangsnetz signifikante Unterschiede bei der Netztopologie, der Netzgröße und den technologischen Randbedingungen gibt.When planning the access area in UMTS networks must a division of the base stations into groups by one radio network controller each can be operated, see above done that the total cost of the system if possible be minimized. However, this planning process is difficult, because different cost contributions such as intra-group connection costs, costs for Radio network controller (RNC) Cost of connecting to a next higher one Hierarchy level and costs for connections between the Clusters in so-called "soft handovers" in a complex manner and sometimes interact in opposite directions. Are there Planning procedure for GSM networks is not easy can be used because it is used, for example, in the UMTS access network significant differences in the network topology, the Network size and the technological boundary conditions there.

Aus [8] ist ein Optimierungsverfahren zur Konfiguration eines hierarchischen Netzwerks bekannt, bei dem eine optimale Knotenzahl ermittelt und eine Neuzuordnung der Knoten in Richtung niedrigerer Hierarchieebenen und dann in Richtung höherer Hierarchieebenen erfolgt, wobei die Optimierung beispielsweise durch Kostenoptimierung erfolgt.From [8] is an optimization method for the configuration of a hierarchical network known in which an optimal Knot count determined and a reallocation of the nodes in Towards lower levels of the hierarchy and then towards higher hierarchy levels takes place, the optimization for example through cost optimization.

Die der Erfindung zugrundeliegende Aufgabe besteht nun darin, ein Verfahren zur Aufteilung eines Telekommunikationsnetzes anzugeben, bei dem eine schnelle und hinsichtlich der Gesamtkosten des Telekommunikationsnetzes möglichst optimale Aufteilung in Gruppen von Basisstationen bzw. in Funknetzwerksubsysteme erfolgt.The object on which the invention is based is now a method for dividing a telecommunications network  specify where a quick and in terms of Total cost of the telecommunications network as optimal as possible Division into groups of base stations or in Radio network subsystems are done.

Diese Aufgabe wird jeweils durch die Merkmale der Patentansprüche 1 oder 3 gelöst. This task is characterized by the characteristics of Claims 1 or 3 solved.  

Die weiteren Ansprüche betreffen vorteilhafte Ausgestaltungen dieser Verfahren.The further claims relate to advantageous refinements of these procedures.

Die Erfindung besteht im Wesentlichen darin, dass mit Hilfe eines Nachbarschaftsgraphen, unter Berücksichtigung aller relevanten Kostenfaktoren für Ausrüstung und Verbindungen, auf schnelle Weise eine möglichst kostenoptimale Aufteilung eines Telekommunikationsnetzes in Cluster von Basisstationen erfolgen kann. Durch die Clustereinteilung wird die Anzahl der Basisstationen im jeweiligen Cluster und die Anzahl der Netzwerksteuereinheiten im Telekommunikationssystem bestimmt.The invention essentially consists in that with the help of a neighborhood graph, taking into account all relevant cost factors for equipment and connections, a quick and cost-effective distribution a telecommunications network in clusters of base stations can be done. By clustering the number the base stations in the respective cluster and the number of Network control units in the telecommunications system determined.

Die Erfindung wird nachfolgend anhand von in den Zeichnungen dargestellten Ausführungsbeispielen näher erläutert.The invention is described below with reference to the drawings illustrated embodiments explained in more detail.

Dabei zeigtIt shows

Fig. 1 eine prinzipielle Topologie eines terrestrischen Radio Access Networks mit zwei RNC-Clustern, FIG. 1 shows a basic topology of a terrestrial radio access networks with two RNC clusters

Fig. 2 eine Kostenfunktion für die Netzwerksteuereinheiten, Fig. 2 is a cost function for the network control units,

Fig. 3 eine Kostenfunktion für die Verbindungen, Fig. 3 is a cost function for the compounds

Fig. 4 eine Darstellung mit der geografischen Verteilung von Basisstationen, Fig. 4 is a representation of the geographical distribution of base stations,

Fig. 5 einen zu Fig. 4 gehörigen Nachbarschaftsgraphen, Figure 5 is a corresponding to Fig. 4 neighborhood graph.,

Fig. 6 einen minimal spannenden Baum zu den Fig. 4 und 5, Fig. 6 shows a minimum spanning tree to FIGS. 4 and 5,

Fig. 7 eine Darstellung der durch eine Aufspaltung des in Fig. 5 dargestellten Baumes erzeugten Cluster und FIG. 7 shows the clusters and clusters generated by splitting the tree shown in FIG. 5

Fig. 8 eine Darstellung von Kostenfunktionen in Abhängigkeit der Anzahl der erzeugten Cluster. Fig. 8 is an illustration of cost functions depending on the number of clusters generated.

Das UMTS-Netzwerk besteht aus zwei Teilen: Dem Core-Netzwerk und dem Radio Access Network. Das terrestrische UMTS Radio Access Network UTRAN verbindet die üblicherweise als Node B bezeichneten Basisstationen mit den Nutzern des Core- Netzwerks.The UMTS network consists of two parts: the core network and the Radio Access Network. The terrestrial UMTS radio Access Network UTRAN connects the usually as Node B designated base stations with the users of the core Network.

In Fig. 1 ist eine prinzipielle Topologie eines solchen UMTS-Radio Access Netzwerkes mit zwei sogenannten RNC- Clustern beispielhaft gezeigt. Es beinhaltet die mit Node B bezeichneten Basisstationen, die Netzwerksteuereinheiten bzw. Radio Network Controller (RNC) und weitere Netzwerkelemente wie ein Mobile Switching Center MSC und einen bedienenden GPRS-Unterstützungsknoten SGSN, der das Radio Access Network UTRAN mit dem Core-Netzwerk verbindet. Die Basisstationen Node B in einer RNC-Region, die als Radio Network Subsystem RNS bezeichnet wird, sind mit dem Radio Network Controller RNC verbunden, welcher selbst wiederum mit dem Mobile Switching Center MSC und dem bedienenden GPRS Supportknoten verbunden ist.In Fig. 1 shows a basic topology of such a UMTS radio access network with two so-called RNC clusters exemplified. It contains the base stations labeled Node B, the network control units or Radio Network Controller (RNC) and other network elements such as a Mobile Switching Center MSC and an operating GPRS support node SGSN, which connects the Radio Access Network UTRAN to the core network. The base stations Node B in an RNC region, which is referred to as the Radio Network Subsystem RNS, are connected to the Radio Network Controller RNC, which in turn is connected to the Mobile Switching Center MSC and the serving GPRS support node.

Das Ziel einer optimierten Planung eines UMTS-Access Networks ist es, abgesehen von der Unterstützung des Netzwerkplaners und einer Beschleunigung des Planungsprozesses, die Minimierung der Gesamtkosten für die Errichtung und den Betrieb des Netzwerkes. Die Kosten bestehen im Wesentlichen aus Ausrüstungskosten und Verbindungskosten.The goal of optimized planning of a UMTS access network it is apart from the support of the network planner and an acceleration of the planning process that Minimize the total cost of construction and construction Operation of the network. The main costs are from equipment costs and connection costs.

Die Ausrüstungskosten beinhalten die Kosten für die Netzwerksteuereinheiten RNC und Kosten für Zugangsmöglichkeiten. In Fig. 2 sind beispielsweise die Kosten CostRNC für die Netzwerksteuereinheiten RNC in Abhängigkeit des maximal möglichen Datenverkehrs Traffic dargestellt.The equipment cost includes the cost of the network control units RNC and the cost of access. In Fig. 2 the cost Cost RNC for the network controllers RNC in response to the maximum possible traffic Traffic example are shown.

Um die Implementierung nur lose an die Hardwarespezifika zu koppeln werden alle Kostenbestandteile durch Kostenfunktionen beschrieben, die stückweise konstant oder linear sind und vom Benutzer definiert werden können. Die Ausrüstungskosten sind eine Funktion von einem Parameter und weisen typischerweise eine Treppenfunktion auf, deren Stufen mit zusätzlich erforderlichen Einschüben in einem gegebenen Rack korrespondieren.In order to implement the hardware loosely will link all cost components through cost functions  described, which are piecewise constant or linear and from Users can be defined. The equipment costs are a function of a parameter and typically have a staircase function, the steps with additional required racks in a given rack correspond.

Die Kosten für die Basisstationen Node B und die weiteren Einrichtungen MSC und SGSN werden hierbei nicht berücksichtigt, da sie als konstant angenommen werden können.The cost of the base stations Node B and the others Institutions MSC and SGSN are not hereby taken into account because they can be assumed to be constant.

In Fig. 1 sind auch die verbindungsabhängigen Kosten Iub, Iur sowie Iu_cs und Iu_ps angedeutet, wobei die Kosten Iub Kosten für gruppeninterne Verbindungen, die Kosten Iur Kosten für Verbindungen zwischen den Gruppen und die Kosten Iu_cs und Iu_ps gruppenexterne Verbindungskosten darstellen.The connection-dependent costs Iub, Iur and Iu_cs and Iu_ps are also indicated in FIG. 1, the costs Iub representing costs for intra-group connections, the costs Iur costs for connections between the groups and the costs Iu_cs and Iu_ps representing external group costs.

Diese Kosten sind in Fig. 3 als Kosten CostL in Abhängigkeit der Verbindungslänge Length und in den maximalen Datenverkehr traffic in einer Art Treppenfunktion dargestellt. Verbindungskosten sind eine Funktion von zwei Parametern, nämlich der Länge der Verbindungsstrecke und dem Datenverkehr auf dieser Verbindungsstrecke. Diese Kosten sind typischerweise stückweise linear in der Dimension der Länge und stückweise konstant hinsichtlich des Datenverkehrs und korrespondieren zum Beispiel zu einem Tarifsystem eines Providers mit gemieteten Verbindungsleitungen. Jedoch können auch Mikrowellenverbindungen, wie unten näher ausgeführt, berücksichtigt werden.These costs are shown in FIG. 3 as costs Cost L as a function of the connection length and in the maximum data traffic traffic in a kind of staircase function. Connection costs are a function of two parameters, namely the length of the link and the data traffic on this link. These costs are typically piece-wise linear in the dimension of length and piece-wise constant with regard to data traffic and correspond, for example, to a tariff system of a provider with rented connecting lines. However, microwave connections, as detailed below, can also be taken into account.

Da Verbindungskosten, beispielsweise auf einer monatlichen Basis verrechnet werden, wohingegen Ausrüstungskosten nur einmal bezahlt werden, ist eine Amortisationsperiode anzusetzen, um beide Kosten vergleichbar zu machen.Because connection costs, for example on a monthly basis Basis will be charged, whereas equipment costs only Paying once is a payback period to make both costs comparable.

Die Ausrüstungskosten und die Verbindungskosten hängen von der Anzahl und Größe der erzeugten RNC-Cluster ab: Falls viele Cluster erzeugt werden, entstehen Kosten für viele kleine Netzwerksteuereinrichtungen RNC, die dazu neigen, teuerer als weniger große RNCs zu sein; und so steigen die Ausrüstungskosten mit dem Anstieg der Clusteranzahl. Auf der anderen Seite, mit großen Gruppen oder Clustern werden längere Verbindungsstrecken innerhalb eines Clusters benötigt und führen so zu höheren Verbindungskosten. Offensichtlich gibt es eine Kompromisslösung für eine bestimmte Anzahl von Gruppen mit einer bestimmten Anzahl von zugeordneten Basisstationen, die die Gesamtkosten unter Einhaltung der technologischen Randbedingungen minimiert.The equipment cost and the connection cost depend on the number and size of the RNC clusters generated: if  many clusters are created, costs arise for many small network control devices RNC that tend to to be more expensive than less large RNCs; and so they rise Equipment costs with the increase in the number of clusters. On the other side, with large groups or clusters longer connection distances within a cluster are required and thus lead to higher connection costs. Obviously there is a compromise solution for a certain number of Groups with a certain number of assigned Base stations that meet the total cost technological constraints minimized.

Die erfindungsgemäßen Verfahren basieren alle auf einem Nachbarschaftsgraphen (proximity graph). Dies trägt der Tatsache Rechnung, dass Basisstationen, die nahe beieinander liegen, zu einer Gruppe RNC zusammengefasst werden, da sowohl die Verbindungskosten innerhalb einer Gruppe als auch die Verbindungskosten für die Übergabe des Datenverkehrs von einer Gruppe zur anderen (soft handover) reduziert werden. Vorteilhafterweise wird hierbei der Nachbarschaftsgraph nur einmal berechnet und ein mit der Anzahl der Basisstationen quadratischer Anstieg der Entfernungsberechnungen vermieden.The methods according to the invention are all based on one Proximity graph. This carries the Fact that base stations that are close to each other lie together in a group RNC because both the connection costs within a group as well as the Connection costs for the transfer of data traffic from one group to another (soft handover). The neighborhood graph is only advantageous here calculated once and one with the number of base stations Avoid quadratic increase in distance calculations.

In Fig. 4 sind eine Vielzahl von Basisstationen Node B in Form von kleinen Dreiecken dargestellt, die entsprechend ihrer geografischen Lage platziert sind und jeweils einen bestimmten Datenverkehr aufweisen.In FIG. 4, a plurality of base stations Node B are shown in the form of small triangles, which are placed according to their geographical location and each having a particular data traffic.

In Fig. 5 ist ein Nachbarschaftsgraph PG dargestellt, der auf der Basis der in Fig. 4 dargestellten Anordnung von Basisstationen gebildet wurde. Hierbei sind die Kanten des Graphen PG durch Verbindungsstriche zwischen den dreieckförmig dargestellten Basisstationen angedeutet. Der Nachbarschaftsgraph beinhaltet also Knoten, die die Basisstationen Node B und Kanten, die Verbindungen von Basisstationen großer geographischer Nähe repräsentieren. Hierbei kann die Nähe durch eine Entfernungsgrenze definiert werden: Knoten sind nah zueinander, falls ihre Entfernung unterhalb einer Schranke ist. Hierbei besteht jedoch das Problem, eine geeignete Schwelle für unterschiedliche Planungsbereiche oder Unterbereiche zu finden. Anstelle dessen werden erfindungsgemäß für jeden Knoten vi seine k nächsten Nachbarn wj ermittelt und Nachbarschaftskanten für diese Paare (vi, wj) gebildet. Um die k nächsten Nachbarn auf effiziente Weise zu ermitteln, werden alle Basisstationen Node B in einer Variante eines kd-Baumes [1], [2] gespeichert, der diese Art von Abfragen unterstützt. FIG. 5 shows a neighborhood graph PG that was formed on the basis of the arrangement of base stations shown in FIG. 4. The edges of the graph PG are indicated by connecting lines between the triangular base stations. The neighborhood graph thus contains nodes that represent the base stations Node B and edges, the connections of base stations of great geographic proximity. The proximity can be defined by a distance limit: nodes are close to each other if their distance is below a barrier. Here, however, there is the problem of finding a suitable threshold for different planning areas or sub-areas. Instead, according to the invention, its k nearest neighbors w j are determined for each node v i and neighborhood edges are formed for these pairs (v i , w j ). In order to determine the k nearest neighbors efficiently, all base stations Node B are stored in a variant of a kd tree [1], [2] that supports this type of query.

Die Definition der Nähe auf diese Weise hat den angenehmen Nebeneffekt, dass der Knotengrad des resultierenden Nachbarschaftsgraphen begrenzt ist, was zu einem nahezu linearen Zusammenhang zwischen der Anzahl der Basisstationen und der Laufzeit des erfindungsgemäßen Verfahrens führt.The definition of closeness in this way has the most pleasant Side effect that the knot degree of the resulting Neighborhood graphs are limited, resulting in an almost linear relationship between the number of base stations and the duration of the method according to the invention.

Der Nachbarschaftsgraph kann ebenfalls sogenannte "Line of Sight Edges" enthalten, vorausgesetzt, dass durch den Anwender oder durch einen vorgeschalteten Prozess eine ungestörte "Line of Sight" festgestellt wird und eine spezielle Kostenfunktion zur Modellierung der Kosten der Mikrowellenausstattung diesen Kanten zugeordnet wird.The neighborhood graph can also be called a "line of Sight Edges ", provided that through the User or through an upstream process undisturbed "Line of Sight" is determined and a special cost function for modeling the costs of Microwave equipment is assigned to these edges.

Nachdem der Nachbarschaftsgraph gebildet wurde, wird eine Aufteilung der Menge der Kanten entsprechend ihrer Länge vorgenommen, wobei kurze Kanten alle Stationen innerhalb einer Gruppe verbinden sollen und lange Kanten zwischen den einzelnen Gruppen verlaufen sollen. Eine solche Aufteilung wird durch zwei Schritte erreicht:
In einem ersten Schritt wird ein, in Fig. 6 beispielhaft dargestellter, minimaler spannender Baum berechnet, bei dem zum Beispiel die minimale Anzahl von Kanten alle Knoten aus allen Kantenuntermengen mit minimalem Gesamtgewicht verbinden muss. Dies kann auf sehr effiziente Weise durch Benutzung der Algorithmen von Prim [3] oder Kruskal [4] erfolgen. Zu diesem Zeitpunkt ist jedoch der Datenverkehr, der durch die Kanten fließt, unbekannt und so wird die geometrische Länge der Kanten als ihr Gewicht verwendet.
After the neighborhood graph has been formed, the number of edges is divided according to their length, with short edges connecting all stations within a group and long edges running between the individual groups. Such a division is achieved in two steps:
In a first step, a minimum exciting tree, as shown in FIG. 6, is calculated, in which, for example, the minimum number of edges has to connect all nodes from all edge subsets with a minimum total weight. This can be done very efficiently using the algorithms from Prim [3] or Kruskal [4]. At this point, however, the traffic flowing through the edges is unknown and so the geometric length of the edges is used as their weight.

In einem zweiten Schritt wird der so erzeugte Baum in Unterbäume zerlegt, die jeweils eine Gruppe repräsentieren. In Fig. 7 sind die aus dem aufgeteilten minimal spannenden Baum von Fig. 6 gebildeten Cluster oder Gruppen RNS mit einer jeweiligen Netzwerksteuereinheit RKC dargestellt. Die Auftrennung eines Baumes mit gewichteten Knoten in eine minimale Anzahl von Unterbäumen kann sehr schnell durch die Verwendung des Algorithmus von Kundu und Misra [5] erfolgen.In a second step, the tree created in this way is broken down into subtrees, each of which represents a group. FIG. 7 shows the clusters or groups RNS formed from the divided, minimally exciting tree of FIG. 6 with a respective network control unit RKC. A tree with weighted nodes can be separated into a minimal number of subtrees very quickly by using the Kundu and Misra algorithm [5].

Es ist nicht schwierig, weitere Bedingungen in den Aufteilungsprozess einzubauen, zum Beispiel die maximal erlaubte Anzahl von Basisstationen oder den maximalen Datenverkehr pro Netzwerksteuereinheit. Der von den Basisstationen herrührende Datenverkehr wird als Gewichtung verwendet und auf diese Weise eine Aufteilung des Baumes in Teilbäume erreicht, deren Gesamtdatenverkehr unterhalb einer bestimmten Schwelle liegt. Falls diese Schwelle klein gewählt wird, entsteht eine kleine Zahl von Gruppen mit billigen Verbindungsstrukturen aber teueren Hardwareeinrichtungen; falls die Schwelle hoch gewählt wird, entstehen wenige große Gruppen mit teueren Verbindungen zwischen diesen Gruppen und geringere Gesamtkosten für die Hardware, also beispielsweise für die Netzwerksteuereinheiten RNC. Um den besten Schwellenwert zu finden, wird der Baum wiederholt mit unterschiedlichen Schwellen in Unterbäume aufgeteilt und für jede Lösung Kostenbeiträge errechnet: RKC-Kosten für Verbindungen innerhalb einer Gruppe, Kosten für soft handover zwischen den Gruppen und gruppenexterne Kosten, die durch die Verbindungen zwischen den Netzwerksteuereinheiten RNC und den weiteren Einheiten MSC bzw. SGSN hervorgerufen werden. Danach wird der Schwellenwert bestimmt, der zu den geringsten Gesamtkosten führt. Im Folgenden ist ein Pseudocode für ein erstes erfindungsgemäßes Verfahren angegeben.
pseudo-code1:
generate proximity graph;
generate minimum spanning tree
on proximity graph;
for different thresholds t{
cut tree into subtrees using threshold t;
calculate total cost;
if (cost < lowest cost seen so far) best_t = t;
}
cut tree into subtrees using threshold best_t;
It is not difficult to incorporate other conditions in the distribution process, for example the maximum number of base stations allowed or the maximum data traffic per network control unit. The data traffic originating from the base stations is used as a weighting and in this way the tree is divided into subtrees whose total data traffic is below a certain threshold. If this threshold is chosen small, a small number of groups with inexpensive connection structures but expensive hardware devices are created; if the threshold is chosen high, there are a few large groups with expensive connections between these groups and lower overall costs for the hardware, for example for the network control units RNC. In order to find the best threshold, the tree is repeatedly divided into subtrees with different thresholds and cost contributions are calculated for each solution: RKC costs for connections within a group, costs for soft handover between the groups and external costs caused by the connections between the groups Network control units RNC and the other units MSC or SGSN are caused. The threshold value that leads to the lowest total costs is then determined. A pseudocode for a first method according to the invention is specified below.
pseudo-code1:
generate proximity graph;
generate minimum spanning tree
on proximity graph;
for different thresholds t {
cut tree into subtrees using threshold t;
calculate total cost;
if (cost <lowest cost seen so far) best_t = t;
}
cut tree into subtrees using threshold best_t;

In Fig. 8 ist eine Darstellung der Kosten Costs über der Anzahl der Netzwerksteuereinheiten RNC bzw. der Radio Network Subsystems RNS dargestellt, wobei die Gesamtkosten total, die Kosten für die Netzwerksteuereinheiten CostRNC, die Kosten Iub für die Verbindungen innerhalb einer Gruppe, die Kosten Iur für die Verbindungen zwischen den Gruppen und die Kosten Iu für die Verbindungen mit den externen Einheiten MSC und SGSN dargestellt sind. In diesem Beispiel ergibt sich ein Gesamtkostenminimum Min bei 15 Gruppen.In FIG. 8 is an illustration of cost Costs to the number of the network controllers RNC and the radio network subsystem RNS shown, the total cost total, the cost of the network control unit cost RNC, the cost Iub for the connections within a group, the cost Iur for the connections between the groups and the costs Iu for the connections with the external units MSC and SGSN are shown. In this example there is a total minimum cost for 15 groups.

Im oben angegebenen Teil wurde die Baumerzeugung und die Zerlegung getrennt durchgeführt. Das Ergebnis führt zu einer einfach zu implementierenden einfachen und schnellen Methode zur Clusterbildung. Das Verfahren wird jedoch etwas komplizierter, wenn die Baumerzeugung und die Zerlegung ineinander verwoben sind und ein ähnlicher Algorithmus wie der von Kruskal Verwendung findet:
Zu Beginn wird jede Basisstation Node B als Cluster aufgefasst. Die internen Verbindungskosten dieses Clusters sind 0 und die Gesamtkosten für die Netzwerksteuereinheiten RNC sind sehr hoch. Nun werden Cluster solange miteinander verschmolzen, als die Verschmelzung von zwei Clustern die Gesamtkosten reduziert. Die Verschmelzung erfolgt in einer "gierigen" Art und Weise: Das Paar von Gruppen, das zur höchsten Kostenreduktion, also zum größten Gewinn führt, wird also immer zuerst verschmolzen. Diese Paare von Gruppen korrespondieren zu den Kanten des Nachbarschaftsgraphen. Kanten und ihre potentiellen Gewinne werden in einer Prioritätsschlange gespeichert, auf der in jeder Iteration eine Kante mit maximalem Gewinn ausgewählt wird. Prinzipiell müsste der Gewinn für jede Kante nach jedem Verschmelzungsschritt neu berechnet werden. Da jedoch die Gewinne dazu neigen, abzunehmen, ist es sicher, die neue Kalkulation zu verschieben bis eine Kante ausgewählt wird. Diese Überlegungen führen zu dem folgenden Pseudocode:
pseudo-code2:
initialize clusters, one cluster per Node B;
generate proximity graph;
for each edge of the proximity graph {
calculate gain when incident clusters
would be merged;
store edge with its gain in priority
queue q.
}
while (edge e = q.del_max()&&
gain of edge < 0){
if (gain of e not valid anymore){
recalculate gain for edge e;
insert e into q;}
else
{
merge clusters that are incident to edge e;
discard now cluster internal edges;
}
}
In the above part, the tree production and the cutting was done separately. The result is an easy-to-implement, simple and quick method of clustering. However, the process becomes a bit more complicated when tree generation and decomposition are interwoven and an algorithm similar to that of Kruskal is used:
At the beginning, each base station Node B is considered a cluster. The internal connection costs of this cluster are 0 and the total costs for the network control units RNC are very high. Now clusters will be merged as long as the merging of two clusters reduces the total costs. The merger takes place in a "greedy" way: The pair of groups that leads to the highest cost reduction, ie the greatest profit, is always merged first. These pairs of groups correspond to the edges of the neighborhood graph. Edges and their potential wins are stored in a priority queue on which an edge with maximum win is selected in each iteration. In principle, the profit for each edge would have to be recalculated after each merging step. However, since profits tend to decrease, it is safe to postpone the new calculation until an edge is selected. These considerations lead to the following pseudocode:
pseudo-code2:
initialize clusters, one cluster per node B;
generate proximity graph;
for each edge of the proximity graph {
calculate gain when incident clusters
would be merged;
store edge with its gain in priority
queue q.
}
while (edge e = q.del_max () &&
gain of edge <0) {
if (gain of e not valid anymore) {
recalculate gain for edge e;
insert e into q;}
else
{
merge clusters that are incident to edge e;
discard now cluster internal edges;
}
}

Im Vergleich zu dem Baumzerteilungsverfahren erzeugt dieses zweite Verfahren normalerweise mehr Gruppen, da die Gruppen zur Größenverdopplung neigen, wenn sie verschmolzen werden, so dass gegen Ende des Verfahrensablaufs die Clustergröße in größeren Schritten erhöht wird, als bei der Baumzerteilung. Dies führt zu einem früheren Verfahrensabbruch und auf diese Weise zu mehr Gruppen. Das Ergebnis beider Verfahren ist jedoch weitgehend übereinstimmend. Compared to the tree division process, this produces second method usually more groups because of the groups tend to double in size when merged, so that towards the end of the process the cluster size in  larger increments than the tree division. This leads to an earlier termination of the procedure and to this Way to more groups. The result of both procedures is however largely consistent.  

LiteraturblattLiteraturblatt

[1] J. L. Bentley: "Multidimensional Binary Search Trees used for Associative Searching", Comm. of the ACM, 18.09.1975, pp. 509-517
[2] U. Lauther: "A Data Structure for Gridless Routing", Proc. 17th
[1] JL Bentley: "Multidimensional Binary Search Trees used for Associative Searching", Comm. of the ACM, 09/18/1975, pp. 509-517
[2] U. Lauther: "A Data Structure for Gridless Routing", Proc. 17 th

Design Automation Conf., 1980, pp. 603-609
[3] R. C. Prim: "Shortest Connection networks and some generalizations", Beil System Tech. J., 36 (1957), pp. 1389-1401
[4] J. B. Kruskal: "On the shortest spanning subtree of a graph and the traveling salesman problem", Proc. Amer. Math. Soc., 7 (1956), pp. 48-50
[5] S. Kundu, J. Misra: "A linear tree partitioning algorithm", SIAM Journal on Computing, 6 (1977), pp. 151-­ 154
[6] U. Lauther: "The C++ Class Library TURBO - A Toolbox for Discrete Optimization", Software@Siemens, Februar 2000, pp. 34-36
[7] K. Holmberg, "Solving the staircase cost facility location problem with decomposition and piecewise linearization", European Journal of Operations Research, 97: 63-74, 1997
[8] WO 02/35871 A2
Design Automation Conf., 1980, pp. 603-609
[3] RC Prim: "Shortest Connection networks and some generalizations", Beil System Tech. J., 36 (1957), pp. 1389-1401
[4] JB Kruskal: "On the shortest spanning subtree of a graph and the traveling salesman problem", Proc. Amer. Math. Soc., 7 (1956), pp. 48-50
[5] S. Kundu, J. Misra: "A linear tree partitioning algorithm", SIAM Journal on Computing, 6 (1977), pp. 151-154
[6] U. Lauther: "The C ++ Class Library TURBO - A Toolbox for Discrete Optimization", Software @ Siemens, February 2000, pp. 34-36
[7] K. Holmberg, "Solving the staircase cost facility location problem with decomposition and piecewise linearization", European Journal of Operations Research, 97: 63-74, 1997
[8] WO 02/35871 A2

Claims (4)

1. Verfahren zur kostenoptimalen Aufteilung eines Telekommunikationsnetzes in Gruppen (RNS) von Basisstationen (Node B), die durch eine jeweilige Netzwerksteuereinheit (RNC) gemeinsam bedient werden, bei dem
  • a) ein Nachbarschaftsgraph (PG) generiert wird, wobei ausgehend von einer jeweiligen Basisstation, die einem Knoten des Nachbarschaftsgraphen entspricht, Verbindungen mit einem bestimmten Datenverkehr (traffic), die den Kanten des Nachbarschaftsgraphen entsprechen, zu k geographisch nächsten Nachbarknoten gebildet werden,
  • b) auf dem Nachbarschaftsgraphen ein minimal spannender Baum (MST) gebildet wird, wobei eine Teilmenge von Kanten des Nachbarschaftsgraphen so bestimmt wird, dass diese Kanten einen Baum bilden und die Summe der Kantengewichte minimal ist, und wobei die Kantengewichte von der geographischen Länge der Kanten abhängen,
  • c) der minimal spannende Baum in Teilbäume aufgetrennt wird, wobei ein Knoten höchstens über t Kanten verbunden ist und wobei ein Teilbaum eine jeweilige Netzwerksteuereinheit mit einer zugehörigen Gruppe von Basisstationen repräsentiert,
  • d) Gesamtkosten (total) des Telekommunikationsnetzes in Abhängigkeit des durch Aufteilung beeinflussten Datenverkehrs ermittelt werden und
  • e) die Schritte c) und d) für verschiedene Werte von t durchgeführt werden und die Aufteilung mit den geringsten Gesamtkosten ausgewählt wird.
1. Method for the cost-optimal division of a telecommunications network into groups (RNS) of base stations (Node B), which are operated jointly by a respective network control unit (RNC), in which
  • a) a neighborhood graph (PG) is generated, starting from a respective base station that corresponds to a node of the neighborhood graph, connections with a specific data traffic (traffic) that correspond to the edges of the neighborhood graph are formed to k geographically nearest neighboring nodes,
  • b) a minimally exciting tree (MST) is formed on the neighborhood graph, a subset of edges of the neighborhood graph being determined such that these edges form a tree and the sum of the edge weights is minimal, and the edge weights are of the geographic length of the edges depend,
  • c) the minimally exciting tree is divided into subtrees, a node being connected at most via t edges and a subtree representing a respective network control unit with an associated group of base stations,
  • d) total costs (total) of the telecommunications network are determined as a function of the data traffic influenced by the division and
  • e) steps c) and d) are carried out for different values of t and the division with the lowest total costs is selected.
2. Verfahren nach Anspruch 1, bei dem die Gesamtkosten Ausrüstungskosten (z. B. CostRNC), die von der Anzahl der Gruppen abhängen, und Verbindungskosten (CostL) umfassen, wobei die Verbindungskosten von der geographischen Entfernung der zu verbindenden Einheiten und vom maximal möglichen Datenverkehr abhängen. 2. The method of claim 1, wherein the total cost comprises equipment costs (e.g. Cost RNC ) which depend on the number of groups and connection costs (Cost L ), the connection costs depending on the geographical distance of the units to be connected and on depend on the maximum possible data traffic. 3. Verfahren zur kostenoptimalen Aufteilung eines Telekommunikationsnetzes in Gruppen (RNS) von Basisstationen (Node B), die durch eine jeweilige Netzwerksteuereinheit (RNC) gemeinsam bedient werden, bei dem
  • a) zunächst jeder Gruppe eine einzige Basisstation zugeordnet wird,
  • b) ein Nachbarschaftsgraph (PG) generiert wird, wobei ausgehend von einer jeweiligen Basisstation, die einem Knoten des Nachbarschaftsgraphen entspricht, Verbindungen mit einem bestimmten Datenverkehr (traffic), die den Kanten des Nachbarschaftsgraphen entsprechen, zu k geographisch nächsten Nachbarknoten gebildet werden,
  • c) für jede Kante des Nachbarschaftsgraphen ein Gewinn errechnet wird, in dem die Gesamtkosten (total) des Telekommunikationsnetzes in Abhängigkeit des durch Aufteilung beeinflussten Datenverkehrs für den Fall ermittelt werden, dass die über die jeweilige Kante verbundenen Gruppen zu einer Gruppe verschmolzen werden, und mit dem Fall verglichen werden, dass diese Gruppen nicht verschmolzen werden,
  • d) die mit der Kante mit dem größten Gewinn verbundenen Gruppen verschmolzen werden, sofern ein positiver Gewinn vorliegt, wobei die Kanten innerhalb der durch Verschmelzung entstandenen Gruppe beseitigt werden, und
  • e) die Schritte c) und d) solange wiederholt werden solange ein positiver Gewinn auftritt.
3. Method for the cost-optimal division of a telecommunications network into groups (RNS) of base stations (Node B), which are operated jointly by a respective network control unit (RNC), in which
  • a) each group is first assigned a single base station,
  • b) a neighborhood graph (PG) is generated, starting from a respective base station which corresponds to a node of the neighborhood graph, connections with a specific data traffic (traffic) which correspond to the edges of the neighborhood graph are formed to k geographically nearest neighboring nodes,
  • c) a profit is calculated for each edge of the neighborhood graph, in which the total costs (total) of the telecommunications network are determined as a function of the data traffic influenced by division in the event that the groups connected via the respective edge are merged into one group, and with be compared to the case that these groups are not merged,
  • (d) the groups associated with the edge with the greatest profit are merged if there is a positive profit, the edges within the group resulting from the merging being eliminated, and
  • e) steps c) and d) are repeated as long as a positive win occurs.
4. Verfahren nach Anspruch 3, bei dem die Gesamtkosten Ausrüstungskosten (z. B. CostRNC), die von der Anzahl der Gruppen abhängen, und Verbindungskosten (CostL) umfassen, wobei die Verbindungskosten von der geographischen Entfernung der zu verbindenden Einheiten und vom maximal möglichen Datenverkehr abhängen.4. The method of claim 3, wherein the total cost comprises equipment costs (e.g. Cost RNC ) which depend on the number of groups and connection costs (Cost L ), the connection costs depending on the geographical distance of the units to be connected and on depend on the maximum possible data traffic.
DE2002139299 2002-08-27 2002-08-27 Telecommunications network distribution method for cost optimization of radio network sub-systems Expired - Fee Related DE10239299C1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
DE2002139299 DE10239299C1 (en) 2002-08-27 2002-08-27 Telecommunications network distribution method for cost optimization of radio network sub-systems

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
DE2002139299 DE10239299C1 (en) 2002-08-27 2002-08-27 Telecommunications network distribution method for cost optimization of radio network sub-systems

Publications (1)

Publication Number Publication Date
DE10239299C1 true DE10239299C1 (en) 2003-12-18

Family

ID=29557881

Family Applications (1)

Application Number Title Priority Date Filing Date
DE2002139299 Expired - Fee Related DE10239299C1 (en) 2002-08-27 2002-08-27 Telecommunications network distribution method for cost optimization of radio network sub-systems

Country Status (1)

Country Link
DE (1) DE10239299C1 (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008014660A1 (en) * 2006-07-26 2008-02-07 Huawei Technologies Co., Ltd A method for communication from nodeb to cn and corresponding system and devices
WO2017066060A1 (en) 2015-10-12 2017-04-20 T-Mobile Usa, Inc. Cellular backhaul coverage algorithms
CN112101638A (en) * 2020-08-27 2020-12-18 华南理工大学 Cooperative optimization method for urban logistics distribution range

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE10008579C1 (en) * 2000-02-24 2001-04-05 Siemens Ag Cost optimising for fixed network topology of radio communications system
WO2002035871A2 (en) * 2000-10-09 2002-05-02 Telefonaktiebolaget L M Ericsson (Publ) Network topologies

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE10008579C1 (en) * 2000-02-24 2001-04-05 Siemens Ag Cost optimising for fixed network topology of radio communications system
WO2002035871A2 (en) * 2000-10-09 2002-05-02 Telefonaktiebolaget L M Ericsson (Publ) Network topologies

Non-Patent Citations (7)

* Cited by examiner, † Cited by third party
Title
BENTLEY, J.L.: Multidimensional Binary Search Trees Used for Associative Searching. In: Communi-cations of the ACM, 1975, Vol. 18,Nr. 9,S.509-517 *
HOLMBERG, K.: Solving the staircase cast facility *
KRUSKAL, J.B.: On The Shortest Spanning Subtree Of A Graph And The Traveling Salesman Problem. In: Proc. Amer. Math. Soc., Feb. 1956, Vol. 7, S. 48-50 *
KUNDU, S., MISRA, J.: A Linear Tree Partitioning Algorithm. In: SIAM Journal on Computing, 1977, Vol. 6, Nr. 1, S. 151-154 *
LAUTHER, U.: A Data Structure for Gridless Rou- ting. In: Proc. 17th Design Automation Conf., 1980, S. 603-609 *
LAUTHER, U.: The C++ Class Library TURBO - A Toolbox for Discrete Optimization. In: Software (at) Siemens, Februar 2000, S. 34-36 *
PRIM, R.C.: Shortest Connection Networks And Same Generalizations. In: The Bell System Tech. Jour- nal, 1957, Vol. 36, Nr. 6, S. 1389-1401 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008014660A1 (en) * 2006-07-26 2008-02-07 Huawei Technologies Co., Ltd A method for communication from nodeb to cn and corresponding system and devices
WO2017066060A1 (en) 2015-10-12 2017-04-20 T-Mobile Usa, Inc. Cellular backhaul coverage algorithms
EP3332573A4 (en) * 2015-10-12 2019-01-09 T-Mobile USA, Inc. Cellular backhaul coverage algorithms
CN112101638A (en) * 2020-08-27 2020-12-18 华南理工大学 Cooperative optimization method for urban logistics distribution range
CN112101638B (en) * 2020-08-27 2022-06-14 华南理工大学 Cooperative optimization method for urban logistics distribution range

Similar Documents

Publication Publication Date Title
EP2027487B1 (en) Calculation method for network-specific factors in a network of reference stations for a satellite-based positioning system
DE202016107141U1 (en) Network stochastic cross-layer optimization to reach the traffic flow target at minimum cost
DE4128568C2 (en) Multi-layer wiring method for wiring modification on the chip for a highly integrated semiconductor circuit
DE10239299C1 (en) Telecommunications network distribution method for cost optimization of radio network sub-systems
DE60319003T2 (en) Method for redistributing objects to arithmetic units
EP1000513B1 (en) Method of producing a network
DE102011012444A1 (en) Synchronizing volume of data memorized on various hosts that are not connected with one another by wireless ad-hoc-network, comprises combining several nodes into superordinate node and calculating node hash value from hash values of nodes
DE19719170C2 (en) Method and tool for establishing a telecommunications network
EP2924589B1 (en) Onboard unit and method for updating geodata therein
EP1526358A1 (en) Method for parcelization of a road network comprising segments and nodes
EP3475819B1 (en) Method for automatically and dynamically assigning the responsibility for tasks to the available computing components in a highly distributed data-processing system
EP1868114B1 (en) Device and method for storing and retrieving objects with multi-dimensionally addressed electronic data
DE4308512A1 (en) Method for routing connections in a communications network
DE19930119C2 (en) Priority management procedures
DE19911957A1 (en) Remote configurable optical communication network
WO2016071083A1 (en) Method for storage and selection of data
EP0562168B1 (en) Method for routing a data-packet to a destination node in a communications system
EP2086257A2 (en) Method for dynamic allocation of resources from a number of cells of a cellular network
EP0927399B1 (en) Computer-aided optimization of the cluster-distribution of nodes from a node graph representing a technical system
EP0927398B1 (en) Optimization of node couples in a given graph showing coupling possibilities, whereby the graph represents a technical system
EP1298839A1 (en) Network performance management
EP1757999B1 (en) Provision of services in a network
DE60204421T2 (en) Method and apparatus for interference based dynamic frequency channel assignment in a telecommunication network
EP4329266A1 (en) Method and system arrangement for geolocation of data sets
WO2023161369A1 (en) Subscription in concurrently active switching devices

Legal Events

Date Code Title Description
8100 Publication of the examined application without publication of unexamined application
8304 Grant after examination procedure
8364 No opposition during term of opposition
8327 Change in the person/name/address of the patent owner

Owner name: NOKIA SIEMENS NETWORKS GMBH & CO.KG, 81541 MUE, DE

8339 Ceased/non-payment of the annual fee