WO2015081990A1

WO2015081990A1 - Handover selection strategy for mitigating the hidden node problem

Info

Publication number: WO2015081990A1
Application number: PCT/EP2013/075417
Authority: WO
Inventors: Stephen Rayment; Alfred Schmidt; Leif Wilhelmsson
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2013-12-03
Filing date: 2013-12-03
Publication date: 2015-06-11

Abstract

This disclosure relates to a wireless communication network and in particular it relates to a method, performed in a network device 10a in a wireless communication network comprising an access point, of selecting a node 20 for handover to an alternate access point 10b as well as to a node 10a adapted to perform the method. According to one aspect, it presents a method, performed in a network device 10a in a contention based wireless communication network comprising an access point to which access point a number of nodes are connected, of selecting a node for handover to an alternate access point. The method comprises detecting among the connected nodes hidden nodes. The method further comprises analysing, for each identified station if there is at least one alternate access point. When there are several stations that have at least one alternate access point then the method further comprises calculating for each station having at least one alternate access point at least one collision metric, and selecting at least one of the stations having at least one alternate access point for handover to an alternate access point based on the calculated collision metric(s).

Description

Handover selection strategy for mitigating the hidden node problem

TECHNICAL FIELD

This disclosure relates to a wireless communication network comprising an access point to which a number of nodes are connected, and in particular it relates to a method, performed in one or more network devices in the wireless communication network, of selecting a node connected to an access point, for handover to an alternate access point. It also relates to a network device adapted to perform the method.

BACKGROUND

As a means to improve the capacity of mobile networks in the future, wireless local area network, WLAN, technology, and possibly other contention based radio access technologies, are intended to be integral parts of small-cell solutions. That is, the WLAN technology will be regarded as just another radio access technology, (RAT), so that handovers can be made to a WLAN without the user noticing that the service is no longer being carried over 3GPP technologies such as e.g. WCDMA or LTE.

IEEE 802.11 is a set of media access control, MAC, and physical layer specifications for implementing WLAN computer communication. Carrier Sense Multiple Access/Collision Avoidance (CSMA/CA) is a contention based medium access mechanism used in the 802.11 standards to allow distributed coordination of the resources among users contending for the medium. In this disclosure, WLAN is referred to as an example of a radio access technologies intended to be an integral part of a future mobile network. CSMA/CA is therefore briefly described. Carrier Sense Multiple Access, CSMA, is a means to share a channel between devices without the need for centralized control or strict timing. One variant of CSMA is CSMA with collision avoidance, CSMA/CA. In CSMA/CA, the channel is sensed before transmission, and in case the channel is busy, the transmission is deferred. CSMA/CA has many desirable properties, one being that it scales relatively well in that the supported data for an individual user degrades softly when the number of users is increased. This is in contrast to many systems where transmission resources are dedicated to individual users. Since the number of transmission resources is limited, dedicating resources to individual users may impose a distinct limit on the number of users that can be supported.

To control access to the medium, CSMA/CA uses inter-frame spaces, IFS, during which a node, here referring to any wireless device in the CSMA network, must refrain from transmitting, and after which the node must sense the medium to determine whether the medium is busy due to a transmission from another node or RF emitter. The 802.11 standard specifies different IFSs to represent different priority levels for the medium access: the shorter the IFS, the higher the priority. For instance, the Short IFS, SIFS, is used for immediate acknowledgement of a data frame, and the Distributed Coordination Function IFS, DIFS, is used to gain access to the medium to transmit data.

A well-known problem of contention based MAC protocols is the problem with hidden nodes. The so- called Hidden Node problem refers to the situation that one node might not hear another node, and thus when listening for a busy medium, the node might not hear that another node is already transmitting, and therefore initiates a transmission that will cause a collision. See figure 1 for a graphical illustration of the Hidden Node problem. In figure 1, two transmitting nodes, STA 1 and STA 2, that are both contending for the medium, and thus listening to the medium, may not hear each other because they are out of range of each other. At the common destination node, here an access point, AP, a collision occurs, since both STA 1 and STA 2 sensed that the medium was not busy.

The hidden node problem may also occur when the two nodes are associated with different access points, APs, and/or when the nodes are operating on different, yet interfering, WLAN channels. When the nodes are associated with different APs, the nodes may or may not be part of the same Extended Service Set (ESS), and the APs may or may not be controlled by the same WLAN controller.

A known mitigation to this problem is to use the equest-To-Send/Clear-To-Send, RTS/CTS, procedure. Figure 2 illustrates a handshake timing diagram in a CSMA/CA system based on RTS/CTS for unicast data. In figure 2, a first node, being a station STA1, wants to send a data frame to a second node, an access point, AP. The first node then sends a request to send, RTS, control frame to the intended receiver, i.e., the second node. If the receiver is ready to receive, it broadcasts a clear to send, CTS, control frame. After receiving the CTS control frame, the sender transmits the data frame. All other nodes that receive the CTS control frame refrain from transmission.

Furthermore, to allow virtual carrier sensing, every data frame may contain the time needed for its transmission including the ACK control frame. Based on this information, other nodes, for example another station, STA2, will maintain a Network Allocation Vector, shown as NAV in figure 2, to determine when they should sense the medium again. Each of the other nodes decrements its NAV counter by the time since the NAV was received, and no access is allowed as long as the value of the NAV counter is above 0. The other nodes will again sense the medium after the times indicated by the NAV and the subsequent DIFS.

In addition, in order to avoid situations where two nodes transmit at the same time leading to a collision, every node needs to wait for the medium to become free and then invoke the back off mechanism. For this, each node selects a random back off interval, which is illustrated by the box in figure 2 that is filled with slanted lines, within [0, CW], where CW is called the contention window and is initialized to a value CWmin. The node decrements the backoff timer every idle time slot until the counter reaches 0 and the node sends the frame. The CWmin is roughly doubled on each collision until it reaches a maximum threshold called CWmax. Since the RTS and CTS control frames are much shorter than a typical data frame, the probability of a collision from a hidden node with the RTS and CTS control frames is much smaller than the probability of a collision with the actual data frame, which will be protected using this RTS/CTS procedure.

However, even though the RTS/CTS protocol is part of the IEEE 802.11 standard, it is typically not used. The reason is the amount of overhead that the RTS/CTS procedure adds to the data transmission. It has been found that it is often more effective to allow some collisions due to hidden nodes than to protect all transmissions with RTS/CTS, as most of the transmissions will not actually need to be protected.

With the closer integration of 3GPP technologies and WLAN, nodes may be moved from operating in one technology or standard to the other. This may be done for a variety of objectives such as load balancing or maximizing the node's quality of service. To achieve this in an effective way, it is necessary to determine the quality of the radio link to which the node may be handed over. For instance, in the case where a node is being considered for handover from 3GPP technology to WLAN technology, the quality could be measured as the bit rate that can be provided to the node in each of the technologies.

In the IEEE Global Communications Conference (GLOBECOM), 2012, paper "REM based approach for hidden node detection and avoidance in cognitive radio networks", by T. Farnham, it is proposed to avoid the hidden node problem by handing the node over from one access point to another access point. In a more complex scenario with many stations, this is implemented as an iterative process to find which of the nodes should be handed over.

SUMMARY

The basic concept of this invention is to address the hidden node problem in by performing a handover of one or more nodes that are causing the hidden node problem from an access point that uses a Radio Access Technology, RAT, that is prone to the hidden node problem to another access point that uses either the same RAT or another RAT. That is, the choice of RAT or access point for a specific node is at least partly based on whether this node will be a hidden from another node, or another node will be hidden from this node, in the source and target RATs. The decision concerning which node to be handed over from the source RAT to another RAT is taken by the access point in the source RAT or by any other processing device that has been supplied with the necessary measurements needed to make the decision. To keep the overall load of both the source and target RATs at a roughly constant level, the handover of one node from the source RAT to the target RAT may trigger the handover of one or more nodes from the target RAT to the source RAT. For example when a WLAN access point hands over a node to 3GPP to solve the hidden node problem, it may turn out that the load in 3GPP will then become too high. One of the nodes previously connected to 3GPP can be moved to WLAN (assuming it will not be a hidden node) to balance the load.

The present disclosure provides a means to solve the hidden node problem by taking advantage of the ability that one or more of the nodes may be moved to another RAT. Because it is important that a suitable set of nodes is moved from WLAN to the other RAT, a number of different approaches for achieving this are disclosed. What is noteworthy is that the performance is improved not only for the moved nodes, but potentially also for those that are not moved, since the WLAN network can be expected to perform better in terms of e.g. throughput, collision rate and delay. The disclosure also covers the case when the handover is made to another WLAN channel instead, which WLAN channel may be served by either the same access point, or a different but nearby access point.

The present disclosure presents a method, performed in a network device in a contention based wireless communication network comprising an access point, to which access point a number of nodes are connected, of selecting a node for handover to an alternate access point. The method comprises detecting hidden nodes among the connected nodes. The method further comprises analysing, for each identified node being detected as a hidden node if there is at least one alternate access point. When there are several nodes that have at least one alternate access point then the method further comprises calculating for each node having at least one alternate access point at least one collision metric, and selecting at least one of the nodes having at least one alternate access point for handover to an alternate access point based on the calculated collision metric(s). This approach provides a more effective way of selecting a node for hand over to another access point in order to mitigate the hidden node problem. The proposed method takes several factors into account to make sure that all nodes are served in an optimised manner after the handover. In contrast to prior art that suggests that all hidden nodes are moved, a more fine grained decision can be taken. According to one aspect, the method further comprises the step of identifying a change in topology of the wireless communication network. By going through the selection procedure every time the topology is changed, high performance is achieved over time. According to one aspect, the method further comprises determining for each node being detected as a hidden node, an impact of the node on the performance of the access point and identifying nodes having an impact above a level. Then the step of analysing, for each node being a hidden node, if there is at least one alternate access point, is only performed for the hidden nodes identified as having an impact.

According to one aspect, the method further comprises the step of handing over the at least one selected node to an alternate access point. According to one aspect, the step of determining for each node being detected as a hidden node an impact of the node on the performance of the access point comprises, analysing for each node being detected as a hidden node the quality of service of a connection between the access point and the node. Hence, the nodes that cause the most problems may be identified.

According to one aspect, the step of determining for each node being detected as a hidden node, an impact of the node on the performance of the access point comprises, analysing for each node being detected as a hidden node the level of traffic activity of the node. The traffic demand of a node is a very important parameter for determining the impact of the node being a hidden node. Hence, a node being hidden to several nodes may be allowed if it only very rarely transmits data.

According to one aspect, the step of analysing, for each identified node, if there is at least one alternate access point, comprises analysing at least one of predicted bit rate, predicted SSI or the cost of a connection between the respective node and an access point within range of the node. Thus, just because there is an alternate access point within range, this access point may not be suitable for different reasons. By taking different contextual parameters into account, traffic may be controlled in a more efficient way. According to one aspect, the selection of at least one of the nodes, having at least one alternate access point for handover, is further based on predicted performance in alternate access points. Then it is made sure that all nodes in the wireless system are well served. Hence, no node is being subject to being disconnected from the current access point without making sure that there is a good alternate connection. This also implies that one has to consider the need of the respective node. According to one aspect, the selection of one of the nodes that may be moved to another access point is further based on the requirements of the respective node.

According to one aspect, the step of analysing, for each identified node, if there is at least one alternate access point, comprises analysing if there is at least one alternate access point using a radio access technology different from the technology of the access point. According to one aspect, the step of analysing, for each identified node, if there is at least one alternate access point, comprises analysing if there is at least one alternate access point using the same radio access technology as the access point. Thus, handovers may be done to access points utilising either the same and different radio access technologies.

According to another aspect, the disclosure relates to one or more network devices, in a contention based wireless communication network comprising an access point to which a number of nodes are connected, the access point being configured for selection of at least one of the nodes for handover to an alternate access point. The network devices comprise processing circuitry adapted: to detect the hidden nodes among the connected nodes; to estimate, for each of the hidden nodes, the impact of the node on the performance of the Basic Service Set, BSS, that the AP controls; to identify the nodes having an impact above a level, and; to analyse, for each identified node, if there is at least one alternate access point. The processing circuitry is further adapted to, when there are several nodes that have at least one alternate access point, then calculate for each node having at least one alternate access point at least one collision metric, and select at least one of the nodes having at least one alternate access point for handover to an alternate access point based on the calculated collision metric(s). According to one aspect, the processing circuitry is further adapted to identify a change in topology of the wireless communication network.

According to one aspect, the processing circuitry is further adapted to hand over the at least one selected node to an alternate access point.

According to another aspect, the disclosure relates to a computer program, comprising computer readable code, which, when run on one or more network devices in a contention based communication network, causes the network devices to perform the method as described above.

With the above description in mind, the object of the present disclosure is to overcome at least some of the disadvantages of known technology as previously described.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 illustrates a wireless communication network where the hidden node problem may occur.

Figure 2 illustrates RTS/CTS handshake timing diagram for collision avoidance in a CSMA/CA system.

Figure 3 illustrates an access point to which a number of nodes are connected.

Figure 4 is a flowchart illustrating method steps executed in an access point according to different aspects of the disclosure. Figure 5 is a block diagram illustrating an access point. DETAILED DESCRIPTION

The inventors have realised that if a handover is made just to solve the hidden node problem, this may result in a worse situation if care is not taken. Hence, in order to optimize performance, there is a need to further consider which node(s) to handover to another access point or network.

The proposed technique is based on the insight that, in practice, the hidden nodes may or may not be a problem. For example, if the nodes that are hidden have very low traffic and/or low QoS requirements, it may not be an issue in practice. In addition, in the process of finding suitable nodes to be handed over to another access point, it is required that the nodes that are handed over will in fact be handed over to a channel which can support the required data rate and QoS. Moreover, viewed from the system to which the handover is performed, whether it is WLAN or another RAT, it is preferred that the node which is handed over can be supported with a minimum of resources.

In this disclosure, effective methods of selecting a node for handover from an access point, AP, to another radio access technology are proposed. An access point in this application refers to a device that allows wireless nodes to connect to a network using e.g. Wi-Fi, or related standards. Hence, the access point is e.g. a WLAN access point, but it may also be a Machine Type Device giving access to other nodes or a base station in a future radio access system.

In this application the term network device is generally used to define a device where the proposed method is implemented. The network device could be an access point or another processing device that communicates with the access point either wirelessly or wired. In principle, the functionality of the proposed method could be distributed between two or more processing devices, where each device has its own complement of processing circuitry and memory and communication interfaces. Due to the real-time nature of some of the low layer measurements of signal quality, the measurements would typically be performed at the processing device collocated with, or nearest to, the access point that has the radio interface. The remaining functionality may be more easily distributed amongst other processing devices.

A node in this disclosure refers to any wireless device being configured for connecting to an access point. In 3GPP, the RAT-specific term for such a device is user equipment, UE, while in 802.11, the RAT- specific term is station. Examples of user equipments or stations are mobile phones, smartphones, tablets or laptop computers and modems. The term node means any wireless device independent of the radio access technology, RAT, that it uses. Embodiments of the present disclosure are, in general, directed to a CSMA/CA system as described above. However, it must be understood that the same principle is applicable in other systems, where the hidden node problem exists.

A contention based network is a network using a contention based access protocol. A contention based access protocol is a communications protocol for operating wireless telecommunication equipment that allows many users to use the same radio channel or resource without pre-coordination. The "listen before talk" operating procedure in IEEE 802.11 is an example of a contention based protocol.

The term "connection" is used generically within this disclosure to encompass the specific terms used in each radio access technology, RAT, for describing the state where the node and access point have exchanged sufficient signaling information that they are ready to start exchanging data frames. For example, in 3GPP, the RAT-specific term is "connection", while in 802.11, the RAT-specific term is "association".

As described above, the hidden node problem is characterized by having two nodes, STA1 and STA2, which are connected either to the same AP, as in figure 1, or to different APs, wherein STA1 cannot hear STA2. It could also be that the problem is symmetric so that STA2 also cannot hear STA1, but this does not have to be the case. Furthermore, STA1 and STA2 may be operating on different WLAN channels that interfere with each other. In addition, one or both of the STAs may be an AP itself.

Figure 3 illustrates an access point 10a to which a number of nodes 20a-20g are connected, wherein the proposed technique for selecting a node for hand over may be implemented. The access point 10a is typically a WLAN access point that uses CSMA/CA. In the example of figure 3, two nodes, 20a and 20g, are hidden to two other nodes, 20b and 20e, because nodes 20b and 20e are out of range of transmissions from nodes 20a and 20g. Hence, nodes 20b and 20e cannot hear transmissions 30a, 30g of the hidden nodes 20a and 20g. In this scenario there are potentially many collisions at the access point 10a. The method, performed in a network device, such as an access point or another processing node associated with the access point, in a contention based wireless communication network of selecting a node 20a for handover to an alternate access point 10b, will now be described referring to figure 4. For simplicity, the examples are sometimes described with only two nodes, as illustrated in figure 1. However, the method is of course applicable for several nodes as well, for example as illustrated in figure 3. In this example the method is performed in the access point 10a. Alternatively, the selection function could be performed at any other computing node in the network, such as a WLAN controller node, as long as the underlying F physical layer measurements that are collected at the location of the access point are relayed back to the node that is performing the selection function.

The method is e.g. initiated when the topology of the network has changed causing the performance of the access point to decrease. In practice there will be changes in the channel conditions due to that different nodes are moving, new nodes are added, old nodes are disconnected, or simply because of movement in the environment. This means that which nodes are hidden from one another is not static but is information that needs to be updated in order to be useful. Hence, according to one aspect, the method is initiated by the optional step SO of identifying a change in topology of the wireless communication network. In one exemplary embodiment, applicable to all aspects below for solving the hidden node problem by performing a handover to another RAT, the AP makes a re-evaluation of the hidden node problem when either a new node is added to the AP or when a presently connected node is disconnected from the AP. The rationale for this is that the topology for the WLAN network obviously has changed and the issue with hidden nodes then may have changed. This embodiment can be implemented such that when a new node connects to an AP, the AP focuses on determining whether this new node potentially is hidden from some of the already connected nodes. It can also be implemented such that if a node disconnects from the AP, and it was known that this node was hidden from another node and/or had some nodes that were hidden from it, one or more nodes may be moved back to WLAN from the other RAT. As an example, suppose that STAl and STA2 of figure 1 are hidden from one another and therefore STAl has been moved to another RAT so that only STA2 is connected to the AP. If STA2 now disconnects from the AP, then STAl may be moved back to WLAN.

In another embodiment, the update concerning potential hidden nodes is based on the detection of a significant variation in received signal strength at the AP, which indicates that the topology of the network might have changed considerably. As an example, suppose STAl and STA2 originally were received with a signal power of -90dBm and that in this case the nodes were hidden from one another. (One can imagine that the nodes were at different sides of the AP and relatively far from the AP as the signal strength is relatively weak.) Next, suppose that the received signal power of STAl increases to, say, -60dBm. This means that node likely has moved so that it is considerably closer to the AP and therefore there is a high probability that it is no longer hidden from STA2 (or that STA2 is hidden from STAl). Therefore, a large change in the received signal strength is used to trigger a re-evaluation of which nodes are hidden from one another. In yet another embodiment, similar to the previous embodiment, rather than considering the received signal strength, a large change in the used modulation and coding scheme (MCS) is used to trigger a re- evaluation. Because a less robust MCS (high data rate) corresponds to favourable channel conditions whereas more robust MCS (low data rate) corresponds to poor channel conditions, a significant change in MCS can be used as an indication that the channel conditions have changed significantly.

Then, the method of selecting a node for handover, starts in step SI, by the access point detecting hidden nodes among the connected nodes, 20a to 20g. Procedures for how to identify that a first node is hidden from a second node is not part of this disclosure, but is addressed in the co-pending patent application PCT/EP2013/062757, filed on 2013-06-19 and entitled "Hidden node counteraction in a wireless network" by Leif Wilhelmsson, Thomas Nilsson, and Anders Furuskar. Whether the information about hidden nodes is obtained from the nodes or if it is determined by the AP itself is not important at this time; the disclosure is applicable irrespectively of this. In the example of figure 3, typically nodes 20a and 20g are identified as hidden nodes, because they cannot be seen from nodes 20b and 20e. As mentioned before this may or may not imply that nodes 20b and 20e are also hidden nodes. It depends on if their transmissions may be heard by the other nodes. For simplicity let us assume that only node 20a and 20g are hidden nodes.

Then the access point analyses S3, for each node being detected as a hidden node, 20a, 20g, if there is at least one alternate access point 10b. In the prior art, it has been assumed that all nodes can be moved to another RAT. In practice, the situation may be such that only a subset of the nodes connected to the AP supports another RAT to which it can make a handover. Hence, the choice of which nodes to move to another RAT is restricted to those supporting another RAT and to which hand over is feasible.

In one embodiment, the network device may prior to analysing, additionally determine S2 for each node being detected as a hidden node 20a, 20g, an impact of the node on the performance of the access point. Then, the step of analysing S3, for each node 20a, 20g being a hidden node, if there is at least one alternate access point 10b, may only be performed for the hidden nodes identified as having an impact. Hence, the hidden nodes may or may not be a problem. For example, if the nodes that are hidden have very low traffic and/or low QoS requirements, it may not be an issue in practice. Therefore, the access point identifies if any of the detected nodes have an impact above a level. Examples of this are presented below.

When there are several nodes that have at least one alternate access point 10b then the access point further calculates S4 for each node having at least one alternate access point 10b at least one collision metric and selects at least one of the nodes 20a, 20g having at least one alternate access point 10b for handover to an alternate access point 10b based on the calculated collision metric(s). In principle this step implies that the access point identifies candidates for moving to another access point and if there are several candidates, the access point tries to estimate to what extent each detected hidden node will affect the performance of the present access point. The access point will try to get rid of the nodes that cause the most problems. For example, hidden nodes that are hidden to many other nodes and that have very high traffic also have risk of collision. According to one aspect of the disclosure, the at least one collision metric comprises the number of nodes to which the hidden node is hidden.

According to one aspect of the disclosure, the at least one collision metric comprises at least one of collision probability and traffic load. Hence, a hidden node having much traffic or load is likely to cause more damage than a less active node.

According to another aspect of the disclosure, the calculation S4 of at least one collision metric is based on the number of corrupted packets per time unit for the respective node. As an example frame error rate is often used as an input to the rate adaptation algorithm that selects the modulation and coding scheme to use for transmissions to a particular destination node. However, when there are frequent collisions, the rate adaptation algorithm may not be able to find a modulation and coding scheme that reduces the frame error rate to the desired level. Hence the amount by which the frame error rate exceeds the usual operating point of the rate adaptation algorithm is a useful measure of the amount of collisions occurring.

For example the access point may compute the number of frames per time unit for each node and compute probability of collision. If a first node transmits 100 packets/second and 10 packets seem to be affected by the hidden node, then the probability of collision is 0.1. If probability is same for all, the traffic load may be used as an additional measure of probability. There are of course other possibilities as well. It must also be understood that several collision metrics may be used singly or in any combination. According to one aspect of the disclosure, the step of selecting S5, at least one of the nodes 20a, 20g having at least one alternate access point 10b for handover to an alternate access point 10b comprises, selecting at least the access point having the highest calculated collision metric(s). This alternative implies e.g. that the hidden node having the highest load is moved to another serving AP. According to one aspect this step may imply that no node is moved. According to one aspect the method further comprises the step of handing over S6 the at least one selected node 20a to an alternate access point 10b. Hence, one or more nodes selected for hand over are now actually handed over to an alternate access technology. According to one exemplary embodiment, once one node has been handed over to another RAT, it is determined whether the problem with a hidden node still exists. If the problem is solved because after moving one node there are no more hidden nodes, or at least no hidden nodes causing any problem, no more nodes need to be moved. For the example above with two nodes that are hidden from one another, this would imply that once one of the nodes is handed over to another RAT, the problem is solved and the second node will not be moved but will remain connected to the AP.

According to one aspect, the step of determining S2 for each node being detected as a hidden node 20a, 20g, an impact of the node 20a, 20g on the performance of the access point 10a comprises, analysing for each node being detected as a hidden node 20a, 20g the quality of service of a connection between the access point 10a and the node 20a, 20g. Quality of service, QoS, is represented by e.g. error rate, bit rate or performance. If QoS is very low, it indicates that the hidden node may be hidden for many other nodes. Thus, high QoS implies high impact on the access point.

According to another aspect, the step of determining S2 for each node being detected as a hidden node 20a, 20g, an impact of the node 20a, 20g on the performance of the access point 10a comprises, analysing for each node being detected as a hidden node 20a, 20g the level of traffic activity of the node 20a, 20g. In this case high activity implies that the hidden node can be expected to have a high impact on the access point.

QoS and traffic activity are just two examples of measures that affect the impact of hidden nodes and they could be used solely or in combination with each other or with other metrics to determine an impact of the node 20a, 20g on the performance of the access point.

According to one aspect of the disclosure, the step of analysing S3, for each node 20a, 20g, being identified as a hidden node, if there is at least one alternate access point 10b, comprises analysing at least one of predicted bit rate, predicted RSSI or cost of a connection between the respective node 20a, 20g and an access point 10b within range of the node 20a, 20g. To determine if there are alternate access points, each node typically detects nodes within its range, e.g. by listening to pilots transmitted by each access point. However, that a node can hear a pilot is not enough in order to determine that an access point is a true alternative. The node or the access point therefore predicts the service level that may be offered by access points in range of each identified node, in order to predict properties of a possible future connection. However, an alternate node is not restricted to meaning an access point to which it is physically possible to establish a connection. According to one example, an access point implying a high cost may not be considered an alternative for that reason. According to one aspect, the selection S5 of one of the nodes 20a, 20g that may be moved to another access point, is further based on the requirements of the respective node. Hence, in order to decide if an alternate access point is feasible comprises analysing e.g. QoS requirements, charging or services of a device or node to be handed over. The method of selecting a node for handover according to any of the preceding claims, wherein the selection S5 of at least one of the nodes 20a, 20g having at least one alternate access point 10b for handover, is further based on a predicted performance in alternate access points 10b.

In yet another embodiment, the previous embodiment is further refined such that performance for the different nodes using the other RAT and/or using WLAN is also taken into consideration when deciding which one(s) of the nodes to move to another RAT. Specifically, when it has been determined that at least one node needs to be moved to another RAT, different options are investigated with a target to obtain as good performance as possible. Returning to the example above with two nodes being hidden from one another or from other nodes, the performance for respective nodes is first determined. As an example, if it is found that two nodes 20a, 20g (of Fig. 3) have similar performance when using WLAN, but the first node 20a would have superior performance in the RAT to which the nodes can be handed over, then the first node 20a rather than the second node 20g will be moved to the other RAT. The two latter embodiments are illustrated with only two nodes for the ease of illustration. As is readily understood, in a typical scenario the number of nodes is considerably higher and the algorithms for determining which one(s) of the node to move to the other RAT will be correspondingly more complex. One or more of a variety of performance metrics could be used for comparing the performance of the nodes that are hidden from each other. First, one or more performance metrics would be used to compare the performance of the nodes in the source base node where each node is currently associated. In one embodiment, the performance metric would be a function of a measurement of received signal strength at the AP of a signal transmitted from the node. In another embodiment, the performance metric would be a function of a measurement of the downlink and/or uplink throughput that the node has had in the access point over a recent time period. In a more elaborate embodiment, the performance metric would be a function of the estimated instantaneous data rate (or MCS) that could be used for transmissions to or from the node and the estimated fraction of the available channel time that the node could use. The channel time estimate could be based on recent measurements of the channel business. In another embodiment, the performance metric would include a function of the recently measured frame success rate to the node. In another embodiment, the performance metric would include an estimate of the recently measured latency in delivering frames to the node. In another embodiment, the performance metric would include the capabilities of the node in terms of support for various optional 802.11 features, such as supported rate set, the supported 802.11 physical layer mode (e.g., the direct sequence spread spectrum (DSSS), extended rate PHY (ERP), Orthogonal frequency division multiplexing (OFDM), high throughput (HT), and/or very high throughput (VHT) physical layers), frame aggregation, number of spatial streams supported for spatial multiplexing or spatial diversity, the RF band(s) and channels that are supported, or RF bandwidth (e.g., 20, 40, 80 or 160 - MHz RF bandwidth support), where in each case the performance metric could be used to indicate a preference of moving the node that supports a less efficient set of 802.11 features.

In another embodiment, the performance metric would include consideration of the administrative domain(s) of a potential target access point, 10b, and whether the node or source access point have preferences for different administrative domains. For example, the preferences could be based on roaming arrangements between the operators of the source and target access points. These preferences could be configured and stored in the node, or alternatively be communicated from the source and/or target access point to the nodes, or alternatively be stored only in the source and/or target access point. It should be noted that the preference information in the access point could equally well be stored in a server elsewhere in the core network and made available only on demand to the access points or nodes needing the information.

If the node is not currently connected to the target radio access technology, RAT, one embodiment of the metric of performance in the target RAT would be estimates of the average per node throughput in the target RAT. This would be useful in cases where the two nodes are being considered for handover to different target RAT cells, or different RATs. If the target RAT cell is the same for both nodes that are being compared, then one embodiment would involve having the node measure the signal quality of the target RAT cell and informing the AP of this measurement. The signal quality measurement could be a measurement of the signal strength and/or the signal to interference plus noise (SINR) measurement received by the node from the target RAT cell.

One or more metrics would be used to compare the performance of the nodes in their target RAT. If the node is simultaneously connected to the target RAT, the metrics could be one of the same performance metrics that were discussed above for use in the source RAT, i.e., received signal strength, recent throughput, instantaneous data rate and channel time, latency, and supported 802.11 features.

According to one aspect, the step of analysing S3, for each identified node 20a, 20g, if there is at least one alternate access point 10b, comprises analysing if there is at least one alternate access point using the same radio access technology as the access point. In other embodiments, corresponding to those above where the handover is described as being to another RAT, the handover may be to a different WLAN channel, where that WLAN channel may be served by either the same AP, or a different but nearby AP. According to one aspect the step of analysing S3, for each identified node 20a, 20g, if there is at least one alternate access point 10b, comprises analysing if there is at least one alternate access point using a radio access technology different from the technology of the access point.

Turning now to figure 5, a schematic diagram illustrating some modules of an exemplary embodiment of a network device 10a being configured for selection of a node 20a for handover to an alternate access point 10b, will be described. In this application the term network device is generally used. The network device could be any network device in the wireless communication network. The network device could be an access point or another processing device as described above. In principle, the functionality could be distributed between 2 or more processing devices, where each device has its own complement of processing circuitry and memory and communication interfaces. Due to the realtime nature of some of the low layer measurements of signal quality, the measurements would typically be performed at the processing device collocated with, or nearest to, the access point that has the radio interface. The remaining functionality may be more easily distributed amongst other processing devices.

In this embodiment the network device is an access point 10a, such as a wireless router, that allows wireless devices to connect to a network. An access point is a device, to which wireless devices, here referred to as nodes, may connect, typically by accessing a shared media.

The network device 10a comprises a controller, CTL, or a processing circuitry or controller (CTL) 11 that may be constituted by any suitable Central Processing Unit, CPU, microcontroller, Digital Signal Processor, DSP, etc. capable of executing computer program code. The computer program may be stored in a memory, MEM 13. The memory 13 can be any combination of a Read And write Memory, RAM, and a Read Only Memory, ROM. The memory 13 may also comprise persistent storage, which, for example, can be any single one or combination of magnetic memory, optical memory, or solid state memory or even remotely mounted memory.

The network device 10a further comprises a communication interface (i/f), 12. When the network device is the access point 10a from which a node is to be handed over, the communication interface is arranged for wireless communication with nodes 20 within range of the access point. The communication interface may then be adapted to communicate over one or several radio access technologies. If several technologies are supported, the access point typically comprises several communication interfaces, e.g. one WLAN communication interface 12a and one cellular communication interface 12b. If the network device is another network processing device, the communication interface is instead adapted to communicate with other network devices in the communication network.

When the above-mentioned computer program code is run in the processing circuitry 11 of the network device 10a, it causes the network device 10a to determine, for each node being detected as a hidden node 20a, 20g, an impact of the node 20a, 20g on the performance of the access point 10a and identify nodes 20a, 20g having an impact above a level.

The processing circuitry is further caused to analyse S3, for each identified node 20a, 20g, if there is at least one alternate access point 10b and when there are several nodes that have at least one alternate access point 10b then the processing circuitry is caused to calculate S4 for each node having at least one alternate access point 10b at least one collision metric, and to select S5 at least one of the nodes 20a, 20g having at least one alternate access point 10b for handover to an alternate access point 10b based on the calculated collision metric(s).

According to one aspect of the disclosure the processing circuitry or controller 11 comprises one or several of: o a first detector 111 configured to detect hidden nodes 20a, 20g among the connected nodes 20a-g, o a determiner 112 configured to determine S2 for each node being detected as a hidden node 20a, 20g, an impact of the node 20a, 20g on the performance of the access point 10a and identifying S2 nodes 20a, 20g having an impact above a level, o an analyser 113 configured to analyse S3, for each identified node 20a, 20g, if there is at least one alternate access point 10b, o a second detector 114 configured to, when there are several nodes that have at least one alternate access point 10b, calculate S4 for each node having at least one alternate access point 10b at least one collision metric, and o a selector 115 configured to select S5 at least one of the nodes 20a, 20g having at least one alternate access point 10b for handover to an alternate access point 10b based on the calculated collision metric(s).

The first detector 111, the determiner 112, the analyser 113, the second detector 114 and the selector 115 are implemented in hardware or in software or in a combination thereof. The modules 111, 112, 113, 114, 115 are according to one aspect implemented as a computer program stored in a memory 13 which run on the processor 11.

According to one aspect the controller is further adapted to identify a change in topology of the wireless communication network. According to one exemplary embodiment this is performed by an identifier 116.

According to one aspect the processing circuitry 11 is further adapted to hand over the at least one selected node 20a to an alternate access point 10b. According to one exemplary embodiment this is performed by a hand over circuit 117.

According to a further aspect the disclosure relates to a computer program, comprising computer readable code which, when run on a network device in a wireless communication network, causes the network device to perform any of the methods described above.

The network device 10a is further configured to implement all the aspects of the disclosure as described in relation to the methods above.

Claims

1. A method, performed in a network device (10a), or in several network devices, in a contention based wireless communication network comprising an access point (10a) to which a number of nodes (20) are connected, of selecting a node (20a) for handover to an alternate access point (10b), the method comprising:

- detecting (SI) among the connected nodes (20) hidden nodes (20a, 20g);

- analysing (S3), for each node (20a, 20g) being a hidden node, if there is at least one alternate access point (10b) and when there are several nodes that have at least one alternate access point (10b) then: · calculating (S4) for each node having at least one alternate access point (10b) at least one collision metric, and

• selecting (S5) at least one of the nodes (20a, 20g) having at least one alternate access point (10b) for handover to an alternate access point (10b) based on the calculated collision metric(s).

2. The method of selecting a node for handover according to claim 1, wherein the method is performed in the access point (10a).

3. The method of selecting a node for handover according to claim 1 or 2, further comprising the step of:

- identifying (SO) a change in topology of the wireless communication network.

4. The method of selecting a node for handover according to any of the preceding claims, further comprising the step of:

- determining (S2) for each node being detected as a hidden node (20a, 20g), an impact of the node(20a, 20g) on the performance of the access point (10a) and identifying (S2) nodes (20a, 20g) having an impact above a level; and wherein the step of analysing (S3), for each node (20a, 20g) being a hidden node, if there is at least one alternate access point (10b), is only performed for the hidden nodes identified as having an impact.

5. The method of selecting a node for handover according to any of the preceding claims, further comprising the step of: - handing over (S6) the at least one selected node (20a) to an alternate access point (10b).

6. The method of selecting a node for handover according to any of claims 4 or 5, wherein the step of determining (S2) for each node being detected as a hidden node (20a, 20g), an impact of the node (20a, 20g) on the performance of the access point (10a) comprises, analysing for each node being detected as a hidden node (20a, 20g) the quality of service of a connection between the access point (10a) and the node (20a, 20g).

7. The method of selecting a node for handover according to any of claims 4 to 6, wherein the step of determining (S2) for each node being detected as a hidden node (20a, 20g), an impact of the node (20a, 20g) on the performance of the access point (10a) comprises, analysing for each node being detected as a hidden node (20a, 20g) the level of traffic activity of the node (20a, 20g).

8. The method of selecting a node for handover according to any of the preceding claims, wherein the step of analysing (S3), for each identified node (20a, 20g), if there is at least one alternate access point (10b) comprises analysing at least one of predicted bit rate, predicted SSI or cost of a connection between the respective node (20a, 20g) and an access points (10b) within range of the node (20a, 20g).

9. The method of selecting a node for handover according to any of the preceding claims, wherein at least one collision metric comprises at least one of collision probability and traffic load.

10. The method of selecting a node for handover according any of the preceding claims, wherein the calculation (S4) of at least one collision metric, is based on the number of frames per time unit for the respective node.

11. The method of selecting a node for handover according to any of the preceding claims, wherein the step of selecting (S5) at least one of the nodes (20a, 20g) having at least one alternate access point (10b) for handover to an alternate access point (10b) comprises, selecting at least the access point having the highest calculated collision metric(s).

12. The method of selecting a node for handover according to any of the preceding claims, wherein the selection (S5) of at least one of the nodes (20a, 20g) having at least one alternate access point (10b) for handover, is further based on predicted performance in alternate access points (10b).

13. The method of selecting a node for handover according to any of the preceding claims, wherein the selection (S5) of one of the nodes (20a, 20g) that may be moved to another access point is further based on the requirements of the respective node.

14. The method of selecting a node for handover according to any of the preceding claims, wherein the step of analysing (S3), for each identified node (20a, 20g), if there is at least one alternate access point (10b) comprises analysing if there is at least one alternate access point using a radio access technology different from the technology of the access point.

15. The method of selecting a node for handover according to any of the preceding claims, wherein the step of analysing (S3), for each identified node (20a, 20g), if there is at least one alternate access point (10b) comprises analysing if there is at least one alternate access point using the same radio access technology as the access point.

16. The method of selecting a node for handover according to any of the preceding claims, wherein the access points uses a Carrier Sense Multiple Access/Collision Avoidance technique.

17. A network device in a contention based wireless communication network comprising an access point to which a number of nodes (20) are connected, being configured for selection of at least one of the nodes (20a) for handover to an alternate access point (10b), the network device (10a) comprising:

- a communication unit (12) and

- processing circuitry (11) adapted to:

• detect among the connected nodes (20) hidden nodes (20a, 20g);

• analyse, for each node (20a, 20g) being a hidden node, if there is at least one alternate access point (10b) and when there are several nodes that have at least one alternate access point (10b) then: o calculate for each node having at least one alternate access point (10b) at least one collision metric, and o select at least one of the nodes (20a, 20g) having at least one alternate access point (10b) for handover to an alternate access point (10b) based on the calculated collision metric(s).

18. The network device (10a) according to claim 17, wherein the processing circuitry (11) is further adapted to:

- identify a change in topology of the wireless communication network.

19. The network device (10a) according to claim 17 or 18, wherein the processing circuitry (11) is further adapted to:

- determine for each node being detected as a hidden node (20a, 20g), an impact of the node(20a, 20g) on the performance of the access point (10a) and identify nodes (20a, 20g) having an impact above a level; wherein the analysis , for each node (20a, 20g) being a hidden node, of if there is at least one alternate access point (10b), is only performed for the hidden nodes identified as having an impact.

20. The network device (10a) according to any of claims 17-19, wherein the processing circuitry (11) is further adapted to:

- hand over the at least one selected node (20a) to an alternate access point (10b).

21. A computer program, comprising computer readable code which, when run on a network device in a contention based communication network, causes the network device to perform the method as claimed in any of claims 1-16.