FI3860221T3 - Method for controlling load distribution in wireless networks with multiple access points - Google Patents

Claims (20)

1 EP3 860 221 Description The present invention relates to a method for load balancing in a communications network and to a computing unit and a computer program product for carrying out the method.

Prior art

Typically, wireless networks comprise multiple access points over a larger area, via which a terminal receives access to the network.

This can be a WLAN network, for example, or a mobile network on the basis of the LTE or 5G standard.

Since each access point has limited capacity, for example in terms of the bandwidth made available for connections, situations may arise in which an access point reaches its maximum capacity and can no longer accept further connections or service requests.

This can lead, for example, to devices that are new to the particular area not being able to establish a connection or to disconnections occurring.

Each access point has a specific coverage area or coverage radius within which a terminal can establish a reliable wireless connection to provide services.

This area is also referred to as a radio cell or cell.

Typically, multiple access points in a network are arranged to provide different overlap areas, in which a terminal is in the coverage area of two or more access points.

This can enable sufficient coverage with drop-free connections for terminals moving from one of these areas to another.

In various cases, this involves handing over an active connection from one access point to another, which is described by the term handover or hand off.

Different variants are possible here; for example, in the case of a soft handover, the connection to the first access point is maintained until the connection to the second access point is fully established, which significantly improves the connection quality in the network.

However, a handover may also be designed as a hard handover without a parallel connection to two access points.

Likewise, a change of radio cell (i.e., of the responsible access point) may also be initiated for a terminal that is logged in to a cell but has no active connection to an access point (cell reselection).

The described changes to another access point for an active connection or an inactive terminal are common in a wide variety of radio standards, for example in

2 EP3 860 221 mobile networks according to 3GPP standards such as GSM, UMTS, LTE, 5G, but also in WLAN networks according to IEEE 802.11 (often referred to there as roaming) and others.

However, the exact design of the implementation, the network elements that decide on a handover, the steps required for the handover of the connection or the cell change and other details depend on the individual standard and are not detailed here.

Furthermore, even handovers of connections between networks of different standards or different radio technologies, for example from 4G to 5G, are possible.

The described difficulty of congested access points can occur in any wireless networks.

In US 2018/0242160 A1, a method for load balancing in a mobile radio system is described, wherein the load of neighboring access points is taken into account.

The direction of radiation and thus the coverage area of the access points may be changed in each case in such a way that the required load levels can be met and a handover to suitable access points may be performed.

DE 102017 108 428 A1 describes a handover method in a wireless network.

This uses known travel routes (for example, data from a vehicle navigation system) to find the best following cell for a mobile station as the mobile station moves through the various cells.

US 2017/0208502 A1 describes a method for load balancing with which the channel utilization between a terminal and an access point is to be optimized and a different radio channel is selected if necessary on the basis of various parameters.

Disclosure of the invention According to the invention, a method for load balancing in a communications network along with an executing computer program and a computing unit having the features of the independent claims are proposed.

The dependent claims and the following description relate to advantageous embodiments.

In particular, this proposes a method for load balancing in a communications network with which a load level of a first access point of the communications network is initially determined.

This load level is determined at least on the basis of parameters of the services provided by the access point to terminals.

A

3 EP3 860 221 service can be understood, for example, as any telecommunications service as in the 3GPP standard.

If the load level of the first access point is above a predetermined first threshold value, the handover of at least one service to a suitable second access point is prepared, wherein the second access point is one of one or more neighboring access points of which the coverage areas at least partially overlap with the coverage area of the first access point and thus form an overlap area.

This allows the load to be more evenly distributed within overlapping coverage areas.

Preferably, the load level may be determined as a normalized load level, namely as a quotient of an absolute load of the first access point and a maximum load of the first access point.

A normalized load metric on the basis of service-based information may be beneficial for time-sensitive communications in a network, for example.

To carry out a handover of services, the method further comprises receiving a message comprising information regarding the load level of one or more neighboring access points.

For example, such a message may be sent by a central unit of the network, or may be sent by each access point to its neighboring access points at regular intervals.

Further, the method comprises detecting mobility data of at least one terminal, wherein the mobility data indicate one or more future positions of the terminal in one or more predetermined time periods.

On the basis of the detected mobility data for the at least one terminal and on the basis of parameters of the services that are anticipated to be provided by the access point to the at least one terminal at the future point in time, the part of the future load of the access point at a future point in time is then determined, and thus, as a sum, also the total load level of the first access point at a future point in time.

By taking into account mobility data indicating expected positions of a terminal, a prediction may thus be made about the required resources at an access point, and load balancing may be performed in a timely manner, so that future services may also be reliably provided for new terminals entering a coverage area.

4 EP3 860 221 In particular, such mobility data may be available if the terminals are, for example, driverless vehicles or other terminals of which the position and movement route are at least partially known on the basis of control data.

For this purpose, the mobility data should also further comprise at least one probability value, which indicates with which probability the terminal is located at one of the one or more positions at the predetermined point in time.

Determining a future load level then further comprises weighting a future load caused by the terminal with the probability value.

In this way, a distinction can be made between positions that the terminal is certain or very likely to take and possible further positions, for example in a route planning of an autonomous vehicle, with which the next driving positions are already specified and later driving routes are predetermined with lower probability.

When preparing a service handover for load balancing, for example, the access point with the lowest current load level may be selected from the neighboring access points as the second access point, and then the actual handover of at least one service to the second access point may be initiated.

This ensures that no handover is made to an access point that is also already close to capacity, and ensures the most even distribution and reliable connection possible.

In a further embodiment, preparing a handover of services may further comprise forming a combined load level in the overlap area from the sum of the load levels of all access points of which the coverage areas at least partially overlap.

Then, a check may be made to see if the combined load level in the overlap area exceeds a predetermined second threshold value, wherein no handover is carried out if the combined load level exceeds the second threshold value.

For example, such checkcan easily prevent services from continuing to be handed over in an area in which all responsible access points are already close to their maximum load level.

Since typically no access point can provide more than its maximum load level, a maximum combined load level indicates that all individual load levels are also close to capacity.

The method steps described above may, for example, be carried out directly by an access point in a communications network or by a suitable control unit of the

EP3 860 221 access point.

In parallel or alternatively, the method may also be carried out by a central control unit in the communications network, which can monitor the load levels of the individual access points.

Important application areas for wireless networks or the invention are not only in

5 personal communication, but also in particular in industrial environments, where networked control and information processing is increasingly used for all areas from production to administration.

In such cases, local or private networks are also often used, for which the same difficulties arise in terms of maximum capacity of access points.

One area in which wireless networks are increasingly being used is driverless transport vehicles (FTF, also automated guided vehicles, AGVs). These may be automated vehicles, which are fully controlled remotely by a control unit by means of instructions, or autonomous vehicles, which take over at least part of the control and route planning independently.

Such vehicles are used, for example, for material transport in an industrial context, such as transporting goods in a warehouse.

By changing its position, such a vehicle needs to communicate through different access points to maintain a continuous network connection, for example, when moving from one coverage area to another, so that handovers and cell changes along the route are also used here.

At the same time, high demands are placed with respect to a reliable connection and low latency.

A computing unit according to the invention, for example a control unit of an access point, is configured, in particular programmatically, to carry out a method according to the invention.

Furthermore, the implementation of a method according to the invention in the form of a computer program or computer program product with program code for carrying out all the method steps is advantageous because it is particularly low- cost, in particular if an executing control unit is also used for further tasks and is therefore present anyway.

Suitable data carriers for providing the computer program are, in particular, magnetic, optical, and electric storage media, such as hard disks, flash memory, EEPROMSs, DVDs, and others.

It is also possible to download a program via computer networks (Internet, Intranet, etc.).

6 EP3 860 221 Further advantages and embodiments of the invention can be found in the description and the accompanying drawings.

Of course, the features mentioned above and those still to be explained below can be used not only in the respectively specified combinations, but also in other combinations or alone, without departing from the scope of the present invention.

The invention is illustrated schematically in the drawing on the basis of exemplary embodiments and is described in detail below with reference to the drawing.

Description of the figure Figure 1 shows a schematic representation of a situation in a wireless network in which embodiments of the invention may be applied: Figure 2 shows an example of the application of embodiments in a situation according to Figure 1; Figure 3 shows a process flow of an exemplary embodiment of a load balancing, Figures 4a and b show different points in time of an embodiment in which mobility data are available for a portion of the terminals; and Figure 5 shows a process flow of an embodiment in which mobility data are available for a portion of the terminals.

Detailed description Figure 1 schematically illustrates a situation in a wireless network in which various embodiments could be applied.

Three access points AP1, AP2 and AP3 are shown, through which a terminal can wirelessly connect to a network 100. In general, these can be any access points; in particular, the wireless access points can also be base stations of a mobile communications network, i.e., a node B, for example.

Suitable terminals can communicate with the actual network or core network via the access points.

Intermediate network elements between the access points AP1, AP2, AP3 and the network 100, which again may be designed differently depending on the

7 EP3 860 221 network, such as exchanges, controllers and others have been omitted here for the sake of simplicity.

Each wireless access point may be assigned a coverage area (radio cell) that describes the local area in which there is sufficient reception to reliably connect to aterminal.

Thus, for each of the access points AP1, AP2, AP3 shown in Figure 1,

there is a different coverage area 110, 120, 130 in a radius around the particular access point, shown as circles in the figure.

To the extent that multiple access points are located in sufficient proximity to each other, such coverage areas can also overlap at least partially.

Thus, with overlapping coverage areas, a terminal located in the overlap area 140 can theoretically establish a wireless connection with any of the associated access points.

This can ensure that a terminal can connect to the corresponding network at any time, even when moving from one coverage area to another, or if an access point fails.

Figure 1 now shows a situation in which three terminals D1, D2 and D3 are located in an overlapping coverage area 140 of all three access points and are all connected to the network 100 via the same first access point AP1. Connection or communication with the network via an access point is represented by the course of the dashed arrows in the figure.

This one-sided distribution can provide for a high load up to the load limits of such access point AP1, as shown by the adjacent schematic load indicator 151 in the drawing.

As soon as another terminal D4 enters the coverage area 110 of the already almost fully utilized access point AP1 and wants to establish a connection, the access point AP1 can no longer provide sufficient resources for services, so that a connection or service request is rejected.

In contrast, the access points AP2 and AP3 are relatively lightly utilized at the same time, as indicated by particular load indicators 152 and 153. Figure 2 shows schematically how an overload situation as in Figure 1 may be avoided according to exemplary embodiments.

The three access points AP1, AP2 and AP3 are again given at spatially different positions and with different but partially overlapping coverage areas 110, 120 and 130. As in the first example,

since the three terminals D1, D2, and D3 are all within the overlap area 140, they may connect to any of the access points AP1, AP2, and AP3. To avoid overloading a single access point, communication flows are therefore actively

8 EP3 860 221 assigned to one of the other access points in the overlap area so that the load is distributed more evenly among the access points and it is ensured at all times that sufficient capacity is also available for further terminals or further service requests.

This may be referred to as load balancing.

Prior to load balancing, the situation shown in Figure 1 can exist in which all three terminals are connected to the AP1 access point.

As a result, in the example shown, the network connection of a first terminal D1 to the core network 100 may then be routed via the first access point AP1, for example, the network connection of the second terminal D2 via the second access point AP2, and the network connection of the third terminal D3 via the third access point AP3. Once again, the communication flows are shown by the course of the dashed arrows between the core network and the terminal via the particular access points.

As shown via the schematic load indicators 251, 252, 253 of the access points, the load is now more evenly distributed among the access points and each access point still has spare capacity for additional connections or service requests.

Thus, if a fourth terminal D4 now enters the coverage area 210 of the first access point, a connection to such access point can be reliably established.

It is possible to determine which communication flows are to be preferentially handed over to other access points, in order to enable both even load distribution and continuous provision of the requested services.

For example, in one embodiment, low priority services may preferably be handed over to another access point.

It may optionally also be accepted that the connection quality for such service is degraded, as long as the minimum requirements for providing the service are met.

tis also optionally possible to consider, for example, temporarily dropping a service or terminating a communication connection if the utilization on all available access points is too high, so that insufficient resources are available even with load balancing.

This decision may also be made on the basis of priority values, for example.

9 EP3 860 221 According to exemplary embodiments, such load balancing may occur on the basis of a relative current load of each access point.

For this purpose, a metric for the load can be defined, which can be based on various service-related information.

Various parameters may be used to define a preferred normalized load metric.

In particular, quality of service (QoS) parameters or descriptors (for example, a TSN flow descriptor), as defined in various mobile radio standards and communication standards, may be used for this purpose.

However, all other suitable service- related or connection-relevant parameters may also be used.

For example, the temporal frequency of the particular load-generating communication flow or service may be used as a parameter, wherein a distinction may be made between isochronous, cyclic and acyclic services, for example.

A priority parameter can indicate with which priority the data must be transmitted.

For example, safety-relevant data may be transmitted to a vehicle with high priority.

The amount of data to be transmitted may also be included in the normalized load, for example, in the form of a number of frames required or another suitable measure of the amount of data.

Furthermore, limit values for the expected delay or latency time of a data packet (delay budget, survival time) may be included.

Other conceivable parameters that may be included in a load metric are fluctuations in runtime (jitter), service availability or loss rates for data packets.

From these and other parameters, the normalized load can be derived by applying various arithmetic and/or statistical operations.

For example, two of the above parameters could be multiplied together.

The actual determination of the load from such parameters and the selection of which parameters to include in the load determination will depend on the particular application, and is therefore not shown in detail here.

The load determined in this way ultimately forms the basis for a load balancing method.

To the extent that a load is specified in the following, it can preferably also be a normalized load.

Figure 3 shows a flow chart of an exemplary process flow for load balancing in a situation as in Figure 2. The coverage areas of three wireless access points again form a common overlap area in which three mobile terminals are located.

At the

10 EP3 860 221 beginning of the method, the AP1 access point manages three services for the three terminals. For this purpose, each access point determines its current normalized load level continuously or at predetermined intervals in step 310 according to the following equation: ln (t) = 29,0 < y(t) < 1 (1) where I(t) is the current load level at the nth access point APn, Ln(t) is the current absolute load at this access point APn, and L7%* is the maximum absolute load at the access point APn. In step 320, it is checked whether the load level of the access point thus determined is below a predetermined value. If it is determined that the load level is below predetermined threshold values, the method may be terminated for this time step and the load level may be calculated again in the next time step according to eguation (1). On the other hand, a high utilization on the particular access point, i.e., a case in which the current absolute load approaches the maximum load or the normalized load level approaches 1, L(t) > LR”, ln (8) > 1 (2), is to preferably trigger a redistribution of at least a portion of the services of such access point to another access point. Optionally, a threshold value that is below the maximum load level of the access point may be specified. For this purpose, an access point may further initially check whether sufficient resources are available within the overlap area, i.e., within the area in which multiple coverage areas of multiple access points overlap, by summing the current loadofthe N access points in step 330 and forming a combined normalized load level lora(t) within the overlap area,

N. Lp( ) lotat (t) = a, 0 < lotat (t) <1 (3),

11 EP3 860 221 where L7itt, is a maximum combined absolute load in the overlap area that can be formed from the sums of the individual maximum loads.

It is understood that, in order to determine the combined load level, each access point is to have information regarding the coverage areas of the access points in the vicinity along with their current load levels.

For this purpose, for example,

neighboring access points may periodically communicate their current load levels directly to other access points, or to an element of the network that then communicates or relays the current load levels of the relevant access points to each individual access point.

The coverage area of an access point does not have to be relayed directly as a parameter, as long as it is clear for each access point which surrounding access points form an overlap area, and to which access points services may thus be handed over.

Alternatively or in addition to a current normalized load level, an access point can also indicate its absolute load values along with its limit values for maximum utilization, so that the evaluating access point or another control unit then determines the normalized load level of the other access point from such data.

The information regarding utilization may be transmitted in separate log messages or together with other data in suitable formats.

If it is then determined in step 340 that there are sufficient resources in total within the overlap area, i.e., a predetermined limit value for the combined load level /totai(t)

inthe overlap area is not exceeded, the overloaded access point may prepare a handover of one or more services to another access point in the overlap area.

The limit value may be at a maximum utilization of the access points or at a lower threshold value as described earlier.

For this purpose, a suitable access point may initially be selected from the neighboring access points in step 350. For example, the selection may be based on selecting the access point in the overlap area that has the lowest current load level.

Preferably, the originating access point not only has information regarding the normalized load level of the surrounding access points, but also regarding the absolute load values, so that it may also be checked whether the load parameters of the service to be originated can be sufficiently met by the receiving access point.

Subsequently, the handover of the service is initiated in step 360, for example by sending corresponding log messages to the terminal concerned

12 EP3 860 221 and/or to other access points (for example, the receiving access point) or control units in the network.

If, on the other hand, it is determined in step 340 based on the combined load level check that the combined load of all access points in the overlap area is heading towards or already is (currently) exceeding (or is anticipated to exceed) the defined load limits, i.e., in the case that Yn=1Ln(t) > Liotat ltotat >1 (4),

a handover of services is typically not initially useful or possible.

This ends the evaluation of the current load level for the time being and is carried out again from the beginning for the next time step.

Optionally, in this case, a handover of services between two or more access points can still take place in order to distribute the load more evenly within the overlap area and thus ensure reliable communication, even if this does not free up any additional resources for further terminals.

Here as well, as shown previously, the particular load level of individual originating and receiving access points can serve as the basis for selecting the access point to which a service is to be handed over.

Alternatively or in addition to the handover of services, it may also be specified that if the combined load in the overlap area is too high to make it possible or reasonable to hand over services to another access point, it is further checked whether one of the existing services of an access point can be dropped.

This may be particularly useful if high-priority services are reguested or anticipated to be requested so that a very low-priority service is temporarily dropped, in order to provide sufficient resources at an access point.

Additional conditions and checks may be specified for handing over services to other access points, for example, in order to prevent frequent changes of access points.

Alternatively, such conditions may be taken into account by suitable selection of the parameters involved in the determination of the load.

In addition to the parameters already mentioned, position information may also be taken into account in a load balancing method in further embodiments.

For example, a position of a terminal may either be determined by the network using

13 EP3 860 221 suitable positioning methods, such as triangulation, or may be notified by another control unit that has such information.

This means, for example, that when selecting a suitable access point for a handover of services, in addition to the minimum load, it is also possible to take into account how close a terminal is to the edge of a single coverage area, i.e., how great its current distance is from each access point.

This can prevent a service from being handed over within the overlap area to an access point of which the relative location means that even a small change in the location of the particular terminal requires a new handover to a different access point.

According to further embodiments, position information that is anticipated to change at a future point in time or in a future time period, i.e., mobility data of terminals, may be taken into account in particular.

For example, statistical data that is available in the network and contain information regarding the probability of a large number of requested services or terminals being expected at an access point at specific points in time may be used for this purpose.

These can be obtained, for example, from stored data of past time periods.

Alternatively or additionally, however, mobility information may also be used to determine load balancing, which is at least partially known, for example, on the basis of control data inside or outside the network.

Figures 4a and 4b show two points in time in such an exemplary method.

Once again, a total of four terminals are shown, wherein mobility information for a future time period is now available for at least one of the terminals.

The remaining three terminals D1 to D4 may be any other terminals about which movement no information is available.

Further, as in Figures 1 and 2, three access points that have different but overlapping coverage areas are shown.

The access points are thus positioned in such a way that areas are formed in which a connection to at least two access points is possible in each case.

The mobility information of the fourth terminal D4 is shown using nodes K1 to K4 and describes current and future anticipated position information of the terminal.

Figure 4a shows the current position of the terminal D4 at the first node K1, which is outside all coverage areas of the access points taken into account here.

The

14 EP3 860 221 terminal may be connected to or logged in to an access point that is not shown, or it may be previously without network access.

However, information regarding the anticipated movement of such terminal in a future time period is now available, which represents that the terminal will move along a route along the nodes K2, K3,

and K4. In the present example, the nodes K2 and K3 are located within the coverage area of the first access point AP1 and do not reach any of the coverage areas of access points AP2 and AP3. The following node K4, on the other hand, is again outside the coverage areas of all three access points considered.

Preferably, specific points in time may be specified for one or more nodes along the route of a terminal at which the terminal D4 will reach the particular node.

The points in time may also be specified by indicating a period of time that will still elapse from the notification until the expected arrival of the terminal at the node.

Alternatively, instead of a point in time, broader time intervals, or probabilities that the terminal will be at that particular node at a particular point in time, may be specified.

As a further alternative, a time-dependent function may be specified for a terminal at a specific position, which function describes such probability and thus has a maximum at the most probable time of stay (or for a longer time period), while the probability drops again prior to and after such time period.

By means of such information, the network can make a prediction that or when resources are anticipated to be required for a connection of the terminal D4 to the access point AP1, namely at least at the points in time when the terminal is at a node that is within the coverage area of the first access point AP1. In particular, such mobility information may already be taken into account for load balancing before the terminal is in the coverage areas of the participating access points.

Optionally, the path distance between the nodes may also be taken into account, for example by interpolating the travel path and the resulting positions of the terminal between the nodes, since the entry into the coverage area may already take place before the terminal D4 arrives at the particular node.

Alternatively, without a special calculation, a time period may be provided in each case prior to the point in time of reaching the first node in the coverage area as a buffer, at which time network resources are preferably to be released for the entering terminal at the latest.

The buffer time period may be selected uniformly for all terminals or depending on the type of terminals.

15 EP3 860 221 Therefore, in addition to the previously described determinations of the load levels of each access point and the combined load level for an overlap area, the expected additional load at an nth access point APn at point in time t and the load due to static terminals (or terminals without known mobility information) at such pointin time t may be combined and normalized again on the basis of the maximum load of the access point, so that an expected future load level at point in time t for the nth access point arises: K) = SEO 0 < KO) (5) Since potentially expected connections are also included in this case, the value for the expected future load level can also be greater than 1; in such a case, it is therefore necessary to hand over services to other access points, drop services, and/or reject reguested services, since the maximum load of the access point is anticipated to be exceeded.

In Figure 4a, corresponding to the situation in Figure 1, a currently high load level of the first access point AP1 is shown as an example, since it manages services for all terminals D1, D2 and D3. In addition, the access point AP1 now has access to mobility information regarding a terminal D4 that has not previously been connected to such access point and is not yet in its coverage area.

Thus, carrying out a load balancing method as described above, taking into account the expected future load according to equation (5), will reveal that, due to the expected entry of the terminal D4 into the coverage area 110 of the first access point AP1, at least a portion of the services are to be initially handed over to another access point.

Optionally, the optimum point in time by which the handover of services is to be completed can also be determined.

It is also possible to use the mobility data for the terminal D4 along the nodes K1 to K4 to predict for how long the additional resources for such terminal are anticipated to have to be provided.

In the exemplary situation shown in Figure 4, the result can therefore be a similar load balancing as previously described for Figure 2. Figure 4b shows a later point in time of the situation in Figure 4a.

In the meantime, the terminal D4 has moved to the node K2 as predicted by the mobility data, and is thus in the coverage area of the access point AP1. As a result of the load

16 EP3 860 221 balancing method described, the services for the terminals D2 and D3 were handed over from the access point AP1 to the access points AP2 and AP3, respectively, so that access point AP1, which previously was loaded according to a limit value, now has sufficient resources to provide services for the terminal D4.

Likewise, the access points AP2 and AP3 may also have received notification of the mobility data of terminal D4, so that they can predict, based on the notified route, that it is anticipated that the terminal D4 will not to enter their coverage area and therefore need not be taken into account in the determination of the load, even for future time periods.

Along with the mobility data for a terminal, in addition to the point in time at which the particular position is taken, values for the anticipated load that the terminal will use at such points in time and/or information regarding the relevance and importance of the requested services may also be provided, for example.

Alternatively, default values predefined by the access point or a control unit for the assumed load of a terminal may be used, and the default values may also be set differently depending on other parameters, for example, on the basis of the type of a terminal or on the basis of other parameters sent with the position information.

In further embodiments, terminals that are logged in to an access point without currently using an active connection to the network via the access point may also be included in the load balancing method.

For example, information regarding periodically requested services of a logged-in terminal may be used to also make statements about a future requested load at a specific access point AP1. If it is highly probable that newly established services of logged-in terminals will exceed the predetermined threshold values for the load level of an access point, one or more of the inactive logged-in terminals may be instructed to dial another cell (cell reselection). For example, the future load due to logged-in inactive terminals can be predicted again using statistical or historical data, such as when services are requested and connections are established from specific terminals at periodic intervals.

Such embodiments may of course be combined with the use of mobility data, wherein both mobility data from terminals with active connectivity as well as mobility data from inactive terminals or without connectivity may be taken into account.

17 EP3 860 221

One use case for an exemplary load balancing method using mobility data is communication in the field of automated or autonomous vehicles, in particular driverless transport vehicles.

Figure 5 shows a portion of a schematic process flow for this example.

The vehicles may each comprise at least one wireless communication unit, which enables communication with the desired networks and corresponds to the terminals generally described thus far.

Information known from the control system of the vehicles may be relayed to the access points or to other elements of the network in a variety of ways and may be used in determining current and future load distribution and load balancing.

Such extended method maybe applied in particular if a large portion of the expected communication flows and services in a network is given by mobile terminals with mobility information at hand, for example in a local WLAN network or 5G network with control of driverless transport vehicles.

A control center that is responsible for controlling the transport vehicles and thus has route information for the individual transport vehicles or their terminals may be present.

The route information may be specified by another unit or by default, and/or may be actively determined and calculated by the control center.

To the extent that vehicles are mentioned in the following, these are to be understood as networked vehicles with corresponding terminals that may establish a connection to the access points.

The route of the vehicle may be specified in different ways.

One possible embodiment uses different nodes along which the route runs, as in the example in Figure 4. The vehicle may move in a straight line between the nodes in real terms; alternatively, the nodes may form an approximation of a vehicle's actual route, and the number of nodes along a route may be arbitrarily specified.

Such a system is described in detail, for example, in the publication VDA 5050 - "Interface for the communication between automated guided vehicles (AGVs) and a master control." Optionally, a distinction may thereby also be made between already specified nodes of a route, which have already been released for the vehicle and/or instructed by corresponding driving commands (base nodes), and anticipated nodes (horizon nodes). The anticipated nodes may be, for example, nodes along which the vehicle will move to complete a job in the absence of congestion or other situations that would require rerouting.

18 EP3 860 221 If a distinction is made between specified and anticipated nodes, the probability for the specified nodes, for example, can be close to 1, while the maximum probability for the anticipated nodes is lower and decreases, for example, depending on the distance or with an increasing number of possible crossroads or route alternatives.

A control center for the transport vehicles, for example a master control center,

may have information related to all current positions of all vehicles, and to the planned routes of all vehicles and thus to future positions of the vehicles.

Therefore, in an environment, if a large portion of traffic comes from mobile terminals with known mobility information, control center data may be included to make load balancing decisions.

In an exemplary situation of this type, the terminals can be considered in two groups, namely driverless transport vehicles (or, more generally, terminals with known mobility information), along with a group of other terminals.

The load Ln staf), caused by the other terminals D1, D2 and D3 (i.e., "static"

terminals, or terminals for which no mobility data are available) is preferably treated and evaluated in step 510 as previously described with respect to Figures 1 to 3. The load caused by the transport vehicles, on the other hand, may be treated differently.

An access point sends a request to the vehicle control center at step 512, requesting a list of all vehicles that will be within the coverage area at a predetermined future point in time (for example, "in 60 min"). This future point in time may be specified as a configuration parameter of the load balancing method, and may also be predetermined the same for all access points or notified to all access points.

In addition, for each transport vehicle AGV i in the list, a probability value p(i) may be specified, which indicates the probability that the i! transport vehicle AGV, will be in the coverage area of the requesting access point at the predetermined point in time.

Now, in order to determine the total load Ls acvs(f) of all transport vehicles (AGVSs) for an access point n at point in time t in step 514, such obtained probability p(n,i) with respect to the access point n may be used as a weighting factor for the load of a single transport vehicle i.

The load Lnacv(t,i) of a single transport vehicle i at

19 EP3 860 221 point in time t may be predetermined by an estimate or a value for the average load of a single vehicle obtained from previous data.

Lnacvs(t) = Zio p(n, i) * Li agy (t, i) (6) This may then be used to determine a future load at the access point n at point in timet, wherein the load caused by the transport vehicle(s) (terminals with mobility data) and the load according to equation (6) caused by N terminals without mobility data is summed: LA (8) = Lystart (t) + La acvs(t) (7) The total future load thus determined from equation (7) on the basis of mobility data may be used accordingly as in equation (5) in order to determine the expected future normalized load level of a access point and to evaluate it in step 520 with respect to the predetermined maximum load limits, and then to perform load balancing between the N access points on this basis as described previously.

The further steps for checking the total load in the overlap area and finding a suitable access point correspond to steps 330 to 360 in Figure 3, preferably using the summed load according to equation (7) for all access points, and are not detailed here.

An access point may also hand over multiple services in order to accomplish load balancing, and may either hand over all services to one access point or variably hand over services to different access points.

For example, an anticipated load level may be calculated after a load is delivered, and depending on whether the anticipated load level is then below the desired threshold value for the load level of the access point, the handover may be initiated or other services may be selected for handover to another access point.

The actual steps and logs that initiate and carry out a handover to another access point, network or cell are system-dependent and inherently known in the subject.

Therefore, if it has been determined in the previously described steps that handover of services to another access point is desired, corresponding log messages may be sent from an access point to a terminal for this purpose, for example, notifying a request for handover to the access point selected in

20 EP3 860 221 accordance with the preceding method.

Likewise, as an intermediate step, each access point can be notified of, for example, a list of possible access points (corresponding to a neighbor list) for a handover, the current load levels of which can then also be announced separately or in the form of such list.

In the preceding examples, the evaluation of the load levels and the decision on load balancing were each carried out primarily by the access points.

Likewise, however, other variants are possible, such as the use of a central control unit in the network (such as the AMF, access and mobility management function, in relation to a 5G network), which then collects service-related information from the connections and uses this to derive current load levels and manage load balancing.

Likewise, combinations of multiple units in the network are conceivable.

Similarly, mobility data from mobile terminals may be provided by different suitable entities, or may be collected from different sources by a control unit in the network, for example, and then relayed to the relevant access points as needed.

Itis understood that the preceding steps may be applied to a wide variety of embodiments for terminals for which the network or other control unit has mobility data, and is not limited to the example of driverless transport vehicles.

Similar situations arise, for example, with autonomous or semi-autonomous networked rail vehicles, with which a reliable and delay-free connection is reguired at all times for safety reasons, for example for route signals.

Likewise, for a navigation system in a connected vehicle with which a user has selected a specific driving route, such a route may be used to evaluate mobility data.

The fact that the vehicle remains individually controllable in this process, depending on the embodiment, and that the user may also decide against the selected route, may again be taken into account by including probability values in the mobility data.

Mobility data may also be obtained for individual users of a mobile terminal with a certain reduced probability of occurrence, for example, on the basis of their past behavior, for example by recognizing regular locations at a specific time of day.