WO2014077753A1

WO2014077753A1 - Node and method for selecting radio network layer

Info

Publication number: WO2014077753A1
Application number: PCT/SE2012/051272
Authority: WO
Inventors: Fredrik Persson; Tomas Nylander; Jari Vikberg
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2012-11-19
Filing date: 2012-11-19
Publication date: 2014-05-22

Abstract

Method for selecting a radio network layer for a wireless device located in an area with multiple overlapping radio network layers. The method is performed in a radio network node of a wireless communication system. The method comprises retrieving (710) information from an analysis of traffic generated by a service session of the wireless device, and identifying (720) the service session based on the retrieved information. The method also comprises selecting (730) a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support|of each of the multiple overlapping radio network layers.

Description

NODE AND METHOD FOR SELECTING RADIO NETWORK LAYER TECHNICAL FIELD

The disclosure relates to heterogeneous networks, and more specifically to a radio network node and a method for selecting a radio network layer for a wireless device located in an area with multiple overlapping radio network layers.

BACKGROUND

Long Term Evolution (LTE) is the fourth-generation mobile communication technologies standard developed within the 3^rd Generation Partnership Project (3GPP) to improve the Universal Mobile Telecommunication System (UMTS) standard to cope with future requirements in terms of improved services such as higher data rates, improved efficiency, and lowered costs. GSM EDGE Radio Access Network (GERAN) is the radio access network in a GSM system. EDGE stands for Enhanced Data Rates for GSM Evolution. The Universal Terrestrial Radio Access Network (UTRAN) is the radio access network of a UMTS and Evolved UTRAN (E-UTRAN) is the radio access network of an LTE system. In a GERAN/UTRAN and an E-UTRAN, a User Equipment (UE) is wirelessly connected to a Radio Base Station (RBS) commonly referred to as a Base Transceiver Station (BTS) in GSM, as a NodeB in UMTS, and as an evolved NodeB (eNodeB) in LTE. An RBS is a general term for a radio network node capable of transmitting radio signals to a UE and receiving signals transmitted by a UE. In GSM, a Base Station Controller (BSC) controls the BTS and is connected to the Core Network (CN). The BSC and the BTS are together called the Base Station System (BSS). In UMTS, a Radio Network Controller (RNC) controls the NodeB, and is, among other things, in charge of management of radio resources in cells for which the RNC is responsible. The RNC and its corresponding NodeBs are called the Radio Network Subsystem (RNS). The RNC is in turn also connected to the CN.

Figure 1 illustrates a conventional radio access network in an LTE system. An eNodeB 101 a with a transmission point 102a serves a UE 103 located within the eNodeB's geographical area of service also called a cell 105a. The eNodeB 101 a manages the radio resources in its cell 105a and is directly connected to the CN (not illustrated). The eNodeB 101 a is also connected via an X2 interface to a neighboring eNodeB 101 b with a transmission point 102b serving another cell 105b. Heterogeneous Networks

In a cellular network there may be areas with high traffic, i.e. with a high concentration of UEs. In such areas it is desirable to deploy additional capacity to keep an acceptable user satisfaction. The additional capacity may be obtained by deployment of additional macro base stations, or by deployment of additional nodes of lower output power covering a smaller area in order to concentrate the capacity boost in a smaller area. There may also be areas with bad coverage in a cellular network where there is a need for coverage extension, and again one way to obtain better coverage is to deploy nodes with low output power to concentrate the coverage boost in a small area. One argument for choosing nodes with lower output power in the above cases is that the impact on an existing macro network can be minimized, as it is a smaller area where the macro network may experience interference. Such low power nodes could for example be pico base stations, micro base stations, or femto base stations. The names pico, micro, and femto mainly indicate the magnitude of their output power and thus their coverage.

Currently there is a strong drive in the industry in the direction towards the use of low power nodes. Some different terms used for this type of network deployments are heterogeneous networks, sometimes also called HetNets, or multi-layer networks. Figure 2 illustrates a macro base station 201 which provides a wide area coverage also called a macro cell. Different kinds of low power nodes are deployed to provide small area capacity and/or coverage. In this example, pico base stations 203 serving pico cells, relays 205 with a certain area of service, and clusters of home base stations also called femto base stations 207 serving femto cells are shown. In Figure 3 a heterogeneous network view is exemplified for the LTE case. S1 - MME interface/reference point is used for control signaling between the eNodeBs comprising both a macro eNodeB 304 and a low power micro eNodeB 305, and the Mobility Management Entity (MME) 31 1 . The user plane data goes via the Serving GateWay (S-GW) 312 on S1 -U interface/reference point. Between eNodeBs 304 and 305 the X2 interface/reference point is used. Not shown in the figure is a transport network between the eNodeBs 304 and 305, which are the radio network nodes in E-UTRAN, and the CN nodes.

Furthermore, a heterogeneous network normally consists of both multiple Radio Access Technologies (RAT) like Wi-Fi, GSM/GPRS/EDGE, UMTS and LTE, and multiple radio network layers. A common deployment scenario is two radio network layers, for example a macro network layer and an underlay pico network layer. However, there are also scenarios with more than two levels of radio network layers, as in the example in Figure 2. Deploying additional nodes to achieve a heterogeneous network with multiple overlapping radio network layers is one way to boost capacity and increase end user experience of a service provided over the network. However, this needs to be combined with traffic and UE steering in order to make the most out of the heterogeneous network. As different radio network layers will give different service performance and/or network resource demands, it is desirable to allow a radio network layer selection for mobility decisions which takes e.g. the different service performance of the radio network layers into account.

SUMMARY

It is therefore an object to address the need for intelligent traffic steering between the radio network layers of a heterogeneous network, by providing the possibility to select the most efficient radio network layer for a wireless device considering service sessions of the wireless device. This object and others are achieved by the method and the radio network node according to the independent claims, and by the embodiments according to the dependent claims. In accordance with a first aspect of embodiments, a method for selecting a radio network layer for a wireless device located in an area with multiple overlapping radio network layers is provided. The method is performed in a radio network node of a wireless communication system. The method comprises retrieving information from an analysis of traffic generated by a service session of the wireless device, and identifying the service session based on the retrieved information. The method also comprises selecting a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers.

In accordance with a second aspect of embodiments, a radio network node of a wireless communication system configured to select a radio network layer for a wireless device located in an area with multiple overlapping radio network layers is provided. The radio network node comprises a processor. The radio network node also comprises a memory storing instructions that, when executed, cause the radio network node to retrieve information from an analysis of traffic generated by a service session of the wireless device, and identify the service session based on the retrieved information. The instructions also cause the radio network node to select a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers.

An advantage of embodiments is that it is possible to select a radio network layer for a UE giving the most efficient network utilization, the best quality of experience, and/or the best service or application performance. A further advantage of embodiments is that network layer is selected considering both resource efficiency and service performance. One example is a highly mobile UE running a real-time service like Voice over IP (VoIP), which would get better service experience in the macro radio network layer even if the radio level conditions would result in selection of the pico radio network layer. Other objects, advantages and features of embodiments will be explained in the following detailed description when considered in conjunction with the accompanying drawings and claims.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a schematic illustration of a radio access network in LTE. Figure 2 is a schematic illustration of a heterogeneous network deployment.

Figure 3 is a schematic illustration of a heterogeneous network view for LTE.

Figure 4 is a block diagram schematically illustrating the functional components according to embodiments.

Figure 5 is a schematic high-level architectural view of a deployment in a radio network node according to embodiments.

Figure 6 is a signaling diagram of a procedure for a use case scenario according to embodiments.

Figure 7a-c are flowcharts illustrating the method in the radio network node according to embodiments. Figure 8 is a block diagram schematically illustrating the radio network node according to embodiments.

DETAILED DESCRIPTION

In the following, different aspects will be described in more detail with references to certain embodiments of the invention and to accompanying drawings. For purposes of explanation and not limitation, specific details are set forth, such as particular scenarios and techniques, in order to provide a thorough understanding of the different embodiments. However, other embodiments that depart from these specific details may also exist.

Moreover, those skilled in the art will appreciate that the functions and means explained herein below may be implemented using software functioning in conjunction with a programmed microprocessor or general purpose computer, and/or using an application specific integrated circuit (ASIC). It will also be appreciated that while embodiments of the invention are primarily described in the form of a method and a node, they may also be embodied in a computer program product as well as in a system comprising a computer processor and a memory coupled to the processor, wherein the memory is encoded with one or more programs that may perform the functions disclosed herein.

Embodiments are hereinafter described in a non-limiting general context in relation to example scenarios, where a radio network node selects a radio network layer for a UE in a heterogeneous network comprising a macro radio network layer overlapping with a pico radio network layer, possibly of different RATs comprising GERAN, UTRAN and E-UTRAN. However, it should be noted that the embodiments may be applied to any heterogeneous network deployment with overlapping radio network layers and for any type of RAT or combination of RATs. A wireless device may be any kind of wireless terminal, such as a UE, a portable computer, or a smartphone. Hereinafter the wireless device will be exemplified by a UE.

A service session is the period of time a user interfaces with an application offering a service. The service session begins when the user accesses the application and ends when the user quits the application. The application is a program designed to perform a function or suite of related functions, i.e. a service of benefit to an end user of a UE. Spotify is an example of a music service providing streamed music, which is provided by the Spotify application. Hereinafter, a service session refers to the service actions or transactions during a period of time. Within one service session the end user may add or drop media components, for example starting with voice and adding video for a video call. In that case it is possible that the service requirements change within the same service session. There may also be parallel service sessions, for example a UE handling a voice session and starting a data session for websurf at the same time but not related to the voice session and thus not within the same service session. The problem of how to select a radio network layer for a UE in an area with multiple overlapping radio network layers which gives the most efficient network utilization and/or the best quality of experience or service/application performance, is addressed by a solution where traffic generated by a service session of the UE is analyzed in order to observe characteristics related to the service session and to identify the service session. The information about the service session of the UE may then be used together with knowledge related to service session support of each of the overlapping radio network layers, to select a radio network layer. Some services may e.g. get better service experience in a macro radio network layer even if the radio level conditions would result in selection of a pico radio network layer. One example would be a highly mobile UE running a real-time service like VoIP.

The radio network layer selection may be based on information about a service session, and characteristics related to the service session such as bandwidth requirements, bandwidth symmetry, bandwidth variations, delay requirements, signaling requirements, and sensitivity to loss and interruptions. The information about the service session may be gained through traffic analysis which is typically performed in the RAN.

A radio network layer is defined as a layer comprising cells of a certain type, where the type of the cells is defined by the output power and carrier frequency of the cells. The radio network layers can be of different RATs, which may be important for the radio network selection as different RATs support different types of applications, services or service sessions due to RAT-specific characteristics.

In embodiments of the invention, a new logical radio network function handling the radio network layer selection is provided, which will hereinafter be called the Network Layer Selection Function (NLSF). The NLSF 401 , schematically illustrated in Figure 4, may be located in an existing radio network node such as the eNodeB in E-UTRAN, or in a new separate radio network node. The NLSF 401 selects the most appropriate radio network layer for a UE considering all RATs and radio network layers within which the UE, and its service session(s) and/or application(s) are covered. The intention is to enable a smart selection of a radio network layer, all RATs included. The selection may in embodiments be based on more than just the knowledge about the identified service session and the service or application behavior. It may e.g. be based on signaling need, predicted mobility need, UE type, channel quality, and actual load, as will be further described below. Knowledge about the present and/or past service sessions or flows used by a UE may be considered. Embodiments of the invention also comprise one or more of the following functional components 402-405, illustrated in Figure 4 where the interfaces between the functional components are denoted A-F:

402. Traffic Analyzer: Identifies service sessions, characteristics related to the service sessions, and traffic characteristics through packet inspection and heuristic analysis.

Packet inspection is used as a common term for an analysis of traffic at different levels, from the IP header classification to the Deep Packet Inspection (DPI). Below, a very brief description of the different levels of inspection and analysis is given: i. IP header classification, also known as 5-tuple inspection: At this level of analysis packets are inspected up to the Internet layer, the so-called 5- tuple comprising Source IP address; Source port; Destination IP address; Destination port; and Protocol which runs on Transport layer, e.g. Transmission Control Protocol (TCP), and User Datagram Protocol (UDP)). IP header classification is useful when traffic from certain traffic domains, e.g. the Internet or Virtual Private Networks (VPNs), shall be treated in a specific way. One example is to give all Internet traffic a certain quality of service treatment such as a priority. Another example is to add a different security protocol to a VPN. ii. Shallow inspection, also known as stateful inspection: At this level a transport level protocol state is analyzed by inspecting the current protocol header such as the TCP, or UDP header. The sequence of TCP header flags like SYN, ACK and FIN flags may e.g. be analyzed to provide the state of the connection. Shallow inspection is useful when link layer algorithms are triggered by sequences of events of higher layer protocol interactions, without the need of knowing what content is carried. One example of use is to decrease the user terminal battery consumption by letting lower layer protocol states follow higher layer protocol layers.

5 iii. DPI: At this level data content on the Application Layer is analysed, e.g. HyperText Transfer Protocol (HTTP) state, and video frame content. One common example where DPI is used is caching, where the HTTP request is analyzed to identify which content to fetch from the cache. Link layer algorithms can also be made to adapt to specific types of content or 10 applications.

A heuristic analysis involves pattern detection or statistical identification methods on Application Layer data. Such an analysis is typically needed for classification of services with encrypted content, or for applications that intentionally tries to avoid identification, such as free voice over IP applications that avoid being blocked.

15 403. Radio Network Load Analyzer: Measures average traffic load in cells of two or more radio network layers of the RAN.

404. UE DataBase (DB): Stores data about UE movements and service usage. The data is used to estimate a probability of a certain speed, location and service usage for a specific UE.

20 405. Position Handler: Keeps track of UE positions.

The NLSF 401 may combine information from one or more of the functions 402-

405, briefly described above, to take decision on which radio network layer to select. Knowledge about all available radio network layers that are allowed to be considered by the NLSF, such as information related to service session support of

25 the radio network layers, may be configured in the NLSF 401 , or possibly built up using e.g. automatic neighbor relation functionality.

In embodiments of the invention, the NLSF 401 may either: a) override the normal mobility management triggers, or b) provide yet another parameter into the mobility management decision tree.

The NLSF 401 does not select each individual cell in the network, but the type of cells that exists with regards to output power and RAT, which thus corresponds to selecting a radio network layer.

In embodiments of the invention, a functional component referred to as a Network Mobility Manager (NMM) 406 manages handovers and idle mode selection between cells and network layers based on conventional triggers and methods, with the addition that the NMM 406 also considers input from the NLSF 401 , according to either of the alternatives a) or b) above.

A radio network layer selection is done per UE, and when the UE is in active mode the selection related to a UE may be re-evaluated when a new service session is set up or torn down. When the UE is in idle mode, the selection is performed when the UE attach to the network. Embodiments of the invention are valid for both idle and active mode selection. For idle mode selection, a dedicated or individual priority is set for a UE, according to standardized signaling procedures, with the addition that the priority is set based on a selection made by the NLSF. A priority table used as standard today, which is controlled by a concept called Subscriber Profile Identity (SPID) controlled by the CN, is thus overridden. The above described method for idle mode selection is thus used to select radio network layer for idle UE cell selection. The selection in the NLSF may be based on data from the UE DB, such as data regarding previously used service sessions of the UE.

For active mode selection, active UEs are changing cell through standardized handover or cell reselection procedures, with the addition that the selection is based on the constraints defined by the input from the NLSF in terms of a selected radio network layer. For alternative a) described above, handovers or cell reselections are restricted to cells of radio network layers selected or prioritized by the NLSF, as long as coverage for the selected radio network layer is available. Alternatively, handovers or cell reselections may be restricted to cells of all network layers except for the one selected or indicated by the NLSF as far as coverage permits. In one example embodiment, an "avoid-small-cells-flag" may be used, as will be further explained in the first example scenario below. Furthermore, when a UE has one or several active service sessions and a new session is set up, the priority from the NLSF may be changed. This means that a handover or cell reselection may be forced if the UE is active in another radio network layer, even though the normal mobility management triggers would not trigger a handover due to present coverage and/or signal strengths. For alternative b) described above, the radio network layer selections or priorities from NLSF are seen as one input among many others to the mobility management decision of the NMM. Other input to the mobility management decision may e.g. be conventional signal strength and load parameters. The radio network layer priority may also be given a certain weight in the final mobility management decision tree.

The functional components 401 -406 that were briefly described above will be more thoroughly described hereinafter.

402. Traffic Analyzer

The Traffic Analyzer function is used for identification of behavior and characteristics of the services sessions or applications and analysis of the traffic flows generated by the service sessions or applications run by a UE. Certain characteristics are identified either from the service itself and a priori knowledge about the named service, or from the packet flow characteristics which may be probed and estimated. This requires some sort of intercept or traffic analysis functionality in the mobile network. The exact deployment and location in the RAN or the CN depends e.g. on the required granularity of information. The level of information to probe includes different protocol layers like the transport protocol layer (IP, TCP) as well as the application layer. One example of such traffic analysis functionality may be DPI, with the addition of heuristic analysis in case of encryption. The Traffic Analyzer function may in principle investigate both control and user plane traffic. The main approach for service detection is to investigate user plane traffic.

Interface A (see Figure 4) conveys information about traffic flow characteristics and services from the Traffic Analyzer function 402 to the NLSF 401 . The output from Traffic Analyzer function 402 may also be stored in the UE DB 404 through Interface D.

One specific example of information retrieved by the Traffic Analyzer function 402 is the type of over-the-top service carried by the general data bearer, such as specific symmetric voice over IP services, asymmetric real-time video sessions, or gaming applications. This information gives input on the sensitivity to interruptions from handovers or cell reselections, and on symmetry in bandwidth requirements for uplink and downlink, and on how the bandwidth is varying over time.

Another example of information retrieval is the case when the DPI is located in the CN instead of in the RAN. The information detected in the CN is then signaled to the RAN, for example as part of the user plane traffic. The needed information could thus also in this case be retrieved, but the search process is slower when DPI is performed in the CN. In the downlink direction the needed information would be available in the RAN together with the user plane data and there is really no issue. However, in the uplink direction the DPI detected information would be available in the RAN first after the user plane data is sent to the CN, DPI is performed and then the information about DPI results is sent back to the RAN. Furthermore, DPI in the CN would mean changes in both the CN and the RAN, and the signaling between CN and RAN would need to be specified. Every change in the signaling would also mean changes in both CN and RAN. Therefore, it is advantageous to keep the DPI in RAN as all the needed changes can be implemented within the RAN in this case.

403. Radio Network Load Analyzer (RNLA)

Legacy radio networks keep track of network load conditions, channel quality etc. This information is commonly used in packet scheduling to ensure the most resource efficient radio transmission. One example is the usage of the Channel Quality Indication (CQI), indicating the channel quality for a specific UE. Resource utilization can then be optimized by avoiding transmission to terminals at time instances with bad channel quality. In embodiments of the invention, the Radio Network Load Analyzer functionality gives input to the decision on which network layer to select. Since the output is not targeting per-packet decisions the output may be some kind of average load values. The average could e.g. be calculated over a relatively short period, in the range of seconds or possibly minutes. The average load values indicate the load state of a certain network layer.

Interface B conveys information about network average load conditions for the different radio network layers to the NLSF 401.

A normal case would be to add new connections to network layers with low or moderate load, and avoid those of very high load. However, knowing also the sensitivity to handovers and normal variations in required bit rates, more advanced decisions can be made. Thus a radio network layer may be selected based on all the available information although it is the most loaded radio network layer.

404. UE DB

By establishing a history UE DB tracking events per UE, it is possible to detect and predict for example mobility patterns or service related habits of different UEs. For example, a UE used for machine communication may only send Short Message Service (SMS) and be stationary. The history UE DB 404 shall typically be located somewhere in the mobile network. Interface C conveys information about past mobility patterns and direction of movement, or service related habits on a per UE level, e.g. upon request from the NLSF 401 . The information retrieved from the Traffic Analyzer 402 and Position Handler 405 functions is stored in the UE DB 404 via Interface D and Interface E respectively.

Data from the UE DB may be used to find the probability of a certain speed, position and service usage of a specific UE. By mapping the information to the deployed cell grid, including different radio network layers and RATs, the most appropriate radio network layer may be selected for the actual service identified e.g. by using DPI.

405. Position Handler

5 The Position Handler is a function keeping track of the position of all UEs. By providing the latest number of detected positions, the speed and the direction of movement of the UE can be estimated. That information is useful to be able to predict and analyze different options in the NLSF 401 . The output may be stored in the UE DB 404 through Interface E.

10 401 . NLSF

The NLSF is a logical function that may combine knowledge from the other functional components described above over interfaces A-C to find the radio network layer that optimizes the combined resource efficiency and end user service quality experience. The NLSF 401 may reside for example in the RNC

15 when the UE is located in UMTS, or in the BSC when the UE is located in the GERAN. When the UE is located in LTE, the function may reside for example in the macro eNodeB, or in a more central location introducing a new logical LTE functionality. Still another possibility is a central function that is shared by all the different RATs and network layers. In this case, this function may either take

20 active part in all decision by being the master function for all decisions. In the example illustrated in Figure 5, the decision is kept in the existing RAN nodes, in this case in the RNC 521. The traffic flows 502 from different radio network layers, which may be in different RATs, are analyzed e.g. using DPI 503 of the Traffic Analyzer function 402 described previously. Information from DPI 503 and from

25 the UE DB 404 is used in the NLSF 401 to select a radio network layer and to assist in mobility management decisions for a UE 51 1 which is in an area where a pico RBS 512 and a macro RBS 510 have overlapping coverage.

Interface F conveys the output from the NLSF 401 , i.e. a decision or an advice on what radio network layer to select for a particular UE to the NMM 406. One option related to what information the NLSF conveys to the NMM, is that the NLSF conveys the identity of a single prioritized radio network layer. Another option is that the NLSF conveys a prioritized list of radio network layers.

406. NMM The NMM is the functional entity that normally manages handovers between cells and radio network layers, and provides e.g. measurement report control, measurement report evaluation and handover or mobility decisions. In embodiments of the invention, the NMM also takes the input from the NLSF 401 , delivered over Interface F, into consideration for mobility decisions. Policies can be applied in NMM to cater for cases where contradictory input is present.

The realization of the mobility decision based on the output from the NLSF 401 can be divided into two modes - idle mode and active mode - as already explained above.

Idle mode Priority based cell reselection is a new feature introduced in 3GPP Rel-8 for the purpose of idle mode Inter-RAT and frequency layer cell reselection. Priority based cell reselection is defined for all three RATs i.e. GERAN, UTRAN and E- UTRAN. The basic principle is that the network provides the UE with the absolute priority information for the serving cell and other frequency layers and RATs together with more traditional cell reselection parameters. The main objects with priority based cell reselection are two-fold: The first object is to enable the network to provide additional information to the UE's cell reselection algorithm to reduce the need for UE measurements. This is achieved with Common Priorities. The second object is to make the additional information UE-specific and is achieved with Dedicated or Individual Priorities. The name Dedicated Priority is used in UTRAN and E-UTRAN specifications and the name Individual Priority is used in GERAN specifications. As the names indicate, the Common Priorities are the same for all UEs and are broadcasted as part of the cell system information. The Dedicated or Individual Priorities can be UE or subscription specific and are provided to the UE using unicast and dedicated signaling. Dedicated or Individual Priorities override the common priorities. Furthermore, the UE is provided with a timer value indicating how long the received priority information is to be considered as valid. However, it is likely that the dedicated priority information is refreshed before timeout of this timer as the UE is likely to perform for example a periodic Location Registrations before timeout.

There are different mechanisms for sending the Dedicated/Individual Priorities information to the UE in the different RATs. In GERAN, the priority information may be sent to the UE in a Channel Release message, a Packet Cell Change Order message, and in a Packet Measurement Order message. In UTRAN, the priority information may be sent to the UE in a RRC UTRAN MOBILITY INFORMATION message, and in E-UTRAN in a RRCConnectionRelease message.

The Dedicated/Individual Priorities may in embodiments of the invention be used to steer individual UEs to different RATs and radio network layers when they attach to the network after having been in idle mode. Conventionally, the RAN node is configured with different combinations of Dedicated/Individual Priorities. The RAN node uses a SPID received from the CN to select one of configured Dedicated/Individual Priority information and forwards this to the UE before it goes into idle mode as described above. In embodiments of the invention, this mechanism is overridden, and the NMM 406 functionality comprises selecting one of the configured Dedicated/Individual Priority information based on the selection of radio network layer made by the NLSF. The conventional use of Dedicated/Individual Priorities is thus modified to make an idle mode UE preferring a radio network layer selected by the NLSF when it makes it cell selection.

Active mode

The basic 3GPP mechanism for mobility is that neighbor cell relations are needed in the RAN nodes for each cell. These can be either configured manually or retrieved automatically. The configured information consists mainly of a physical cell identity information to be able to identify the cell from the measurements reports received from the UEs, handover routing information to be able to trigger a handover towards the correct target cell and RAT, and handover algorithm information such as the different signal levels and thresholds and how fast to trigger the handover. The conventional procedure is that the RAN node configures the UEs for measurements reporting, and that the UE responds with the downlink measurements reports indicating the requested information. The RAN nodes may also measure uplink quality. This information, together with the information configured in the RAN node, is used to decide when to trigger handover and which target cell and RAT to select. In embodiments of the invention this procedure is modified, and the NMM functionality in the RAN selects when to trigger handover and which target cell/RAT to select based on the selection of radio network layer made by the NLSF. A modified handover procedure is thus in this example used to move UEs to the radio network layer selected by the NLSF.

Typically, both the NLSF and the NMM functionality are implemented in the same radio network node. However, optionally they may be implemented in different nodes. NMM is typically implemented in the eNodeB in LTE, and NLSF can either be also in the eNodeB, or in another more central location.

In an alternative solution to letting the NLSF make the radio network layer selection, the information related to service session support of each radio network layer is only known by the NMM function. The NLSF will thus instead inform NMM about the service session knowledge, such that the NMM can make the final selection of a radio network layer of the UE.

Examples of use case scenarios

In a first example scenario, a UE in active mode has started a VoIP service session with a delay sensitive codec. The VoIP application and codec, including codec rate, are identified by DPI. The service is in itself symmetric and therefore no specific selection must be done for the uplink and the downlink directions, respectively. Since the VoIP codec is delay and jitter sensitive, the macro cell network layer is selected by the NLSF to avoid interruptions and/or jitter related to different handover procedures. To make this visible to the NMM mechanisms a flag is set, indicating that small cells should be avoided whenever possible. When the UE enters the coverage area of a pico cell, the NMM decides, based on the flag that was earlier set, to keep the session on the macro cell layer. When the VoIP session is ended, and if there is no other session active that would require the flag to be set, the flag is reset. At this stage, when the UE enters the coverage area of a pico cell, the NMM may decide to trigger a handover of the UE from the macro cell layer to the pico cell layer.

If the UE starts another service session which requires the same treatment as VoIP while it is in the pico cell layer, the flag is set again. However, nothing further is done with respect to handover until the UE is leaving the coverage area of the pico cell layer. Only at that moment a macro cell is selected as the preferred alternative to avoid additional handovers.

A service session requiring the same treatment as a VoIP service session may e.g. be a real-time conversational video service session. For the real-time conversational video service there is a constraint in play-out buffer size due to the nature of real-time communication. Delay variations may cause data to come too late so that the video quality degrades. When late data comes to the buffer it cannot be used as it is outdated. Another example is the example of live streaming. The requirements are lower than for real-time communication but stricter than for download. Delay variations may also make the video stall if the buffer runs empty. If it is live content, handover interruptions are worse than a lower bit rate.

Since the above described NMM decisions are based on information on all ongoing service sessions, this first use case assumes that we do more or less continuous DPI. However, another principle may be to not rely on continuous DPI but instead trigger the activation of the DPI based on some specific events. One example would be to not have continuous DPI running while the UE is in active mode in the macro network layer. Instead the DPI would be triggered once the UE reports the detection of for example a pico network layer. The benefit with this approach is that the DPI processing is not needed constantly. However, a drawback may be that performing the DPI only after a specific event has occurred may delay the availability of the needed information.

In a second example scenario exemplifying an idle mode scenario, the NLSF will at the end of a service session of a UE, i.e. some time before the UE enters idle mode, get information from the UE DB 404 about historical UE movements and services used over time. Based on that knowledge, the NLSF selects a radio network layer. The NMM 406 may then, based on the selection, decide what dedicated priorities that should be sent to the UE, i.e. how to steer the UEs idle mode selection of different radio network layers. In one example, the UE mainly uses VoIP and is constantly moving, and the NLSF therefore selects the macro network layer for this UE. In another example, the UE normally downloads videos, and is stationary during these sessions. The NLSF will therefore select a small cell radio network layer with high capacity for this UE.

A more detailed third example scenario is described below, with reference to a sequence flow shown in Figure 6. The example scenario refers to the functional components described with reference to Figure 4. In the example a UE 650 moves at moderate speed, passing cells of different sizes in different radio network layers. The user of the UE 650 starts downloading a file. During the file download, an Internet telephony or VoIP session is started. The download of the large file is using the File Transfer Protocol (FTP) 601 , and the FTP download session is identified in 602 by the Traffic Analyzer function 402. The detected information is fed in 604 to the UE DB 404 over interface D, as well as in 603 to the NLSF 401 over interface A.

The Radio Network Load Analyzer (RNLA) 403 detects in 605 the average load of all cells and network layers in the surrounding of the UE 650, and reports in 606 to the NLSF over interface B. In this example two radio network layer selections are possible: selecting large macro cells of wide coverage but with limited data rates especially closer to the cell edges, or selecting much smaller pico cells which instead can provide much higher data rates within its very limited coverage.

The Position Handler function 405 registers or detects 607 the movement of the UE 650, which it reports in 608 to the UE DB 404 over interface E. The NLSF 401 requests information in 609 from the UE DB 404 over interface C, and gets the latest few positions in 610. From the information in 610 the NLSF 401 may identify the direction and speed of the UE 650. The NLSF 401 combines in 61 1 the information retrieved about the traffic, network load and UE movement in 603, 606 and 610 to make a selection of a radio network layer suitable for this UE. In this case the NLSF 401 identifies that the UE 650 is moving right across both macro and pico cell coverage at moderate speed. Considering the lack of real-time requirements and the potentially large amount of data that can be transmitted as bulk, the pico network layer is selected as it then combines network efficiency with service quality. The decision is signaled to the NMM 406 in 612, which uses the information in its mobility decisions (not illustrated here).

While still downloading the file, the subscriber of the UE starts an VoIP service session 613. As earlier the Traffic Analyzer function 402 identifies in 614 the kind of service and reports to the UE DB 404 and the NLSF 401 that both service sessions are ongoing. All steps and signaling 603-612 described above may then be repeated. The real-time requirements of the VoIP service means that many successive handover interruptions may be very costly quality-wise especially in case seamless handover is not supported, which may be the case between several types of network layers. The amount of handover signaling may become significant compared to the low bit rate of the voice service, and the macro network layer is thus selected as the most optimized choice in NLSF. In a conventional solution, the same radio network layer would have been kept as long as coverage would permit rather than making a handover from pico to macro network layer when the VoIP service is started. Description of method and node with reference to Figures 7a-c and 8

Figure 7a is a flowchart illustrating a method for selecting a radio network layer for a wireless device located in an area with multiple overlapping radio network layers. The method is performed in a radio network node of a wireless communication system. The radio network node may e.g. be a BSC in GERAN, an RNC in UTRAN, or an eNodeB in E-UTRAN. The method comprises: - 710: Retrieving information from an analysis of traffic generated by a service session of the wireless device. The retrieved information may comprise characteristics of a flow of traffic generated by the service session. The analysis of the traffic may be performed by use of at least one of a packet inspection such as DPI or shallow inspection, and a heuristic analysis. The analysis may be performed by the Traffic Analyzer 402 described previously and may be forwarded to the NLSF 401 over interface A.

- 720: Identifying the service session based on the retrieved information. - 730: Selecting a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers.

Figure 7b is a flowchart illustrating the method for selecting a radio network layer for a wireless device located in an area with multiple overlapping radio network layers, according to one embodiment of the invention. The method comprises:

- 710: Retrieving information from an analysis of traffic generated by a service session of the wireless device. The retrieved information may comprise characteristics of a flow of traffic generated by the service session. The analysis of the traffic may be performed by use of at least one of a packet inspection such as DPI or shallow inspection, and a heuristic analysis. The analysis may be performed by the Traffic Analyzer function 402 described previously and may be forwarded to the NLSF 401 over interface A. - 720: Identifying the service session based on the retrieved information.

- 740: Determining an average traffic load in cells of at least two of the multiple overlapping radio network layers respectively. This may be done by the RNLA function 403 described previously and may be forwarded to the NLSF 401 over interface B. - 730: Selecting a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers. The selection is in this embodiment based also on the determined average traffic load.

Figure 7c is a flowchart illustrating the method for selecting a radio network layer for a wireless device located in an area with multiple overlapping radio network layers, according to another embodiment of the invention. The method comprises:

- 710: Retrieving information from an analysis of traffic generated by a service session of the wireless device, as already described above. The analysis may be performed by the Traffic Analyzer function 402 described previously and may be forwarded to the NLSF 401 over interface A.

- 720: Identifying the service session based on the retrieved information.

- 750: Repeatedly retrieving positions of the wireless device. The above described Position Handler function 405 may be used to keep track of the position of all UEs.

- 751 : Storing the retrieved positions of the wireless device. The output from the Position Handler function may e.g. be stored in the UE DB 404 through Interface E. - 752: Estimating at least one of a speed and a direction of movement of the wireless device based on the stored retrieved positions. By using the latest number of detected positions in the UE DB, the speed and the direction of the UE can be estimated. That information may be forwarded to the NLSF 401 over interface C, and the NLSF may perform the estimation. - 730: Selecting a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers. The selection is in this embodiment based also on the estimation of the at least one of the speed and the direction of movement.

- 760: Supporting a mobility decision for the wireless device based on the selected radio network layer.

In a first embodiment, the retrieved information from an analysis of traffic, in 710, may comprise stored information from an analysis of traffic generated by a previous service session of the wireless device. Historical information is thus in this first embodiment retrieved for example from the UE DB 404. In a second embodiment, which may be an alternative or an addition to the first embodiment, the retrieved information may comprise information from an analysis of traffic generated by a currently on-going service session of the wireless device. When the first and the second embodiments are used in combination, the historical information could be used to give a hint about what services to start looking for in an on-going service session, which may speed up the traffic analysis.

As already described above, there are alternative embodiments to how the mobility decision for the wireless device is supported in 760, which are described hereinafter in embodiment A-C:

A. The mobility decision may be a cell selection decision for an idle mode UE which attaches to the network and need to select a cell. Supporting the cell selection decision may then comprise: - Selecting a dedicated priority configuration for the wireless device based on the selected radio network layer.

- Transmitting the selected dedicated priority configuration to the wireless device when releasing a connection to the wireless device.

The UE is thus configured with a dedicated priority which is used for the cell selection. In this way the UE is steered to the radio network layer selected by the radio network node, e.g. in the NLSF 401 .

B. The method mobility decision may be a handover decision for an active mode UE. Supporting the handover decision may then comprise basing the handover decision on the selected radio network layer.

C. In embodiments of the invention, the radio network node performing the mobility decisions such as the handover decisions may be another node than the node performing the radio network layer selection. In such a case, supporting the handover decision comprises forwarding information related to the selected radio network layer to a further radio network node controlling handover decisions of the wireless device, such that the further radio network node can base the handover decision on the forwarded information.

A radio network node 800 of a wireless communication system configured to select a radio network layer for a wireless device located in an area with multiple overlapping radio network layers is schematically illustrated in the block diagram in Figure 8. The radio network node may e.g. be a BSC in GERAN, a RNC in UTRAN, and an eNodeB in E-UTRAN. The radio network node comprises a processor 801 , and a memory 802 storing instructions that, when executed, cause the radio network node to retrieve information from an analysis of traffic generated by a service session of the wireless device, identify the service session based on the retrieved information, and select a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers. The analysis of the traffic may be performed by use of at least one of a packet inspection, and a heuristic analysis.

The memory 802 may also store further instructions that when executed cause the radio network node to retrieve information comprising characteristics of a flow of traffic generated by the service session. Furthermore, the memory 802 may store further instructions that when executed cause the radio network node to determine an average traffic load in cells of at least two of the multiple overlapping radio network layers respectively, and to select the radio network layer based also on the determined average traffic load.

In another embodiment, the memory 802 may store further instructions that when executed cause the radio network node to repeatedly retrieve positions of the wireless device, store the retrieved positions of the wireless device, estimate at least one of a speed and a direction of movement of the wireless device based on the stored retrieved positions, and select the radio network layer based also on the estimation of the at least one of the speed and the direction of movement. In one embodiment, the memory 802 may store further instructions that when executed cause the radio network node to retrieve stored information from an analysis of traffic generated by a previous service session of the wireless device. Alternatively or additionally, the memory 802 may store further instructions that when executed cause the radio network node to retrieve information from an analysis of traffic generated by a currently on-going service session of the wireless device.

The memory 802 may in another embodiment store further instructions that when executed cause the radio network node to support a mobility decision for the wireless device based on the selected radio network layer. The following alternative embodiments can be forseen:

A. The radio network node may comprise a communication unit 803. When the mobility decision is a cell selection decision, the memory 802 may store further instructions that when executed cause the radio network node to support the cell selection decision by selecting a dedicated priority configuration for the wireless device based on the selected radio network layer. The communication unit 803 may be configured to transmit the selected dedicated priority configuration to the wireless device when releasing a connection to the wireless device. If the radio network node is an eNodeB in E-UTRAN, the communication unit may be a transceiver configured to communicate wirelessly via antennas with the wireless device.

B. When the mobility decision is a handover decision, the memory 802 may store further instructions that when executed cause the radio network node to support the handover decision by basing the handover decision on the selected radio network layer. C. When the mobility decision is a handover decision, the memory 802 may store further instructions that when executed cause the radio network node to support the handover decision by forwarding information related to the selected radio network layer to a further radio network node controlling handover decisions of the wireless device, such that the further radio network node can base the handover decision on the forwarded information.

The above mentioned and described embodiments are only given as examples and should not be limiting. Other solutions, uses, objectives, and functions within the scope of the accompanying patent claims may be possible.

Claims

A method for selecting a radio network layer for a wireless device located in an area with multiple overlapping radio network layers, the method being performed in a radio network node of a wireless communication system, wherein the method comprises:

- retrieving (710) information from an analysis of traffic generated by a service session of the wireless device,

- identifying (720) the service session based on the retrieved information, and

- selecting (730) a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers.

The method according to claim 1 , wherein the retrieved information comprises characteristics of a flow of traffic generated by the service session.

The method according to any of the preceding claims, wherein the analysis of the traffic is performed by use of at least one of a packet inspection, and a heuristic analysis.

The method according to any of the preceding claims, further comprising:

- determining (740) an average traffic load in cells of at least two of the multiple overlapping radio network layers respectively,

and wherein the selection (730) is based also on the determined average traffic load.

The method according to any of the preceding claims, further comprising:

- repeatedly retrieving (750) positions of the wireless device,

- storing (751 ) the retrieved positions of the wireless device,

- estimating (752) at least one of a speed and a direction of movement of the wireless device based on the stored retrieved positions, and wherein the selection (730) is based also on the estimation of the at least one of the speed and the direction of movement.

6. The method according to any of the preceding claims, wherein the retrieved information comprises stored information from an analysis of traffic generated by a previous service session of the wireless device.

7. The method according to any of the preceding claims, wherein the retrieved information comprises information from an analysis of traffic generated by a currently on-going service session of the wireless device.

8. The method according to any of the preceding claims, further comprising:

- supporting (760) a mobility decision for the wireless device based on the selected radio network layer.

9. The method according to claim 8, wherein the mobility decision is a cell selection decision, and wherein supporting (760) the cell selection decision comprises:

- selecting a dedicated priority configuration for the wireless device based on the selected radio network layer, and

10. The method according to claim 8, wherein the mobility decision is a handover decision, and wherein supporting (760) the handover decision comprises:

- basing the handover decision on the selected radio network layer.

1 1 . The method according to claim 8, wherein the mobility decision is a handover decision, and wherein supporting (760) the handover decision comprises:

- forwarding information related to the selected radio network layer to a further radio network node controlling handover decisions of the wireless device, such that the further radio network node can base the handover decision on the forwarded information.

12. A radio network node (800) of a wireless communication system configured to select a radio network layer for a wireless device located in an area with multiple overlapping radio network layers, the radio network node comprising a processor (801 ), and a memory (802) storing instructions that, when executed, cause the radio network node to:

- retrieve information from an analysis of traffic generated by a service session of the wireless device,

- identify the service session based on the retrieved information, and

- select a radio network layer among the multiple overlapping radio network layers based on the identified service session and information related to service session support of each of the multiple overlapping radio network layers.

13. The radio network node according to claim 12, wherein the memory (802) stores further instructions that when executed cause the radio network node to retrieve information comprising characteristics of a flow of traffic generated by the service session.

14. The radio network node according to any of claims 12-13, wherein the analysis of the traffic is performed by use of at least one of a packet inspection, and a heuristic analysis.

15. The radio network node according to any of claims 12-14, wherein the memory (802) stores further instructions that when executed cause the radio network node to determine an average traffic load in cells of at least two of the multiple overlapping radio network layers respectively, and to select the radio network layer based also on the determined average traffic load.

16. The radio network node according to any of claims 12-15, wherein the memory (802) stores further instructions that when executed cause the radio network node to:

- repeatedly retrieve positions of the wireless device,

- store the retrieved positions of the wireless device,

- estimate at least one of a speed and a direction of movement of the wireless device based on the stored retrieved positions, and

- select the radio network layer based also on the estimation of the at least one of the speed and the direction of movement.

17. The radio network node according to any of claims 12-16, wherein the memory (802) stores further instructions that when executed cause the radio network node to retrieve stored information from an analysis of traffic generated by a previous service session of the wireless device.

18. The radio network node according to any of claims 12-17, wherein the memory (802) stores further instructions that when executed cause the radio network node to retrieve information from an analysis of traffic generated by a currently on-going service session of the wireless device.

19. The radio network node according to any of the preceding claims, wherein the memory (802) stores further instructions that when executed cause the radio network node to support a mobility decision for the wireless device based on the selected radio network layer.

20. The radio network node according to claim 19 further comprising a communication unit (803), wherein the mobility decision is a cell selection decision and wherein the memory (802) stores further instructions that when executed cause the radio network node to support the cell selection decision by selecting a dedicated priority configuration for the wireless device based on the selected radio network layer, and wherein the communication unit (803) is configured to transmit the selected dedicated priority configuration to the wireless device when releasing a connection to the wireless device.

21 . The radio network node according to claim 19, wherein the mobility decision is a handover decision, and wherein the memory (802) stores further instructions that when executed cause the radio network node to support the handover decision by basing the handover decision on the selected radio network layer.

22. The radio network node according to claim 19, wherein the mobility decision is a handover decision, and wherein the memory (802) stores further instructions that when executed cause the radio network node to support the handover decision by forwarding information related to the selected radio network layer to a further radio network node controlling handover decisions of the wireless device, such that the further radio network node can base the handover decision on the forwarded information.