DE10118501A1 - Optimizing channel allocation between base station and mobile stations, by adaptively varying threshold so that channel blocking and/or channel loss probability is optimized - Google Patents

Abstract

The signal-to-noise ratio (C/I) of a received signal is increased by a predetermined threshold value (t), and the selection of channels for allocation is limited to the set of those available channels that satisfy this extended selection criterion. The threshold value is adaptively varied in predetermined intervals so that the channel blocking probability and/or channel loss probability is optimized.

Description

The invention relates to a method for optimized Channel allocation between a base station and a plurality of mobile stations in a radio system, especially in one cellular mobile radio system, with a respective Possible signal-to-noise ratio of a received signal Interference signals as a selection criterion for a number of channels Channel allocation is used.

Such procedures are called DCA (Dynamic Chan nel allocation) and generally refer to a channel Allocation for mobile radio systems (e.g. Digital European Cord less Telecommunications DECT for cordless phones or Universal Mobile Telecommunication System UMTS) if this a connection between a mobile station and a base establish station. In DCA systems, this changes from Depending on which channels are available are etc. In other systems, the channels are technically fi xiert, d. that is, the device transmits only on one or on certain frequencies.

The need for such channel allocation arises for example in cellular mobile networks on the one hand Establishing a connection (call request) or in the event of a transfer (Handover request) of an existing connection from a Ba sis station to the next when changing the mobile station from egg one cell to the next.

The duration of such an operation lies in today's techni systems at about 5 to 10 ms. Especially in Nordame rika is also the term handoff instead of handover bräuchlich. The so-called handover activity is one  essential factor for network activity and thus for Network performance.

There are a variety of solutions within the DCA strategies approaches known, including the first-available approach, according to which always the first free channel out of a set of for one Zu available channels.

In this context it is also important which Un Subset of channels provided for allocation will and how the quality and other parameters of these channels appearance. In this regard, so far are not exclusively self-adaptive methods become known.

The object of the present invention is to use radio systems Make dynamic channel allocation more efficient. Insbesonde Based on the DCA principle, re should make better use of it the radio spectrum resource with a lower probability of loss and less handover achieved with the same traffic or equivalent more traffic with the same loss probability likely to be worn.

According to the present invention, this object is achieved resolved that the process described above continued is formed by the respective signal-to-noise ratio by ei a predetermined threshold is increased and the selection of Channels for channel allocation based on the amount of such available Channels is restricted that this extended selection criteria terium are sufficient, with the threshold in predefinable time are adaptively varied such that a channel Blocking probability and / or channel loss probability probability to be optimal.

According to the invention there is an additional threshold (threshold) introduced, being an essential finding is that network performance in terms of  Blocking and loss probability as well as handover very much depends on this threshold.

Advantageous embodiments of the invention vary this Threshold value permanent and independent (selfadaptive), always looking for optimal network performance. there serve, for example, to prevent blocking and loss probability as the primary criterion.

The threshold value is optimized according to the following process steps:

• a) specification of an initial threshold value,
• b) Determine the channel blocking probability and the Channel loss probability as a function of this Threshold values over a predefinable evaluation period,
• c) incrementing the threshold value by a predeterminable Increment
• d) Determine the channel blocking probability and the Channel loss probability as a function of this current threshold value via the specifiable evaluation period and determining an associated current cost function based on the determined values,
• e) Determination of the local slope from the canal Blocking probability function and from the channel Loss probability function based on the determined Values as well as the subsequent estimation of the respective radio channel blocking probability and channel Loss probability for a fictional to the given Level increment incremented threshold on the one hand and for a fictional one to decrement the given step size threshold to the other,
• f) Determining the incremental and fictitious of the fictitious decremented threshold associated fictitious costs functions based on the estimated values,
• G)
  • - if the fictitious incremental threshold value associated fictitious cost function is smaller than the current cost function determined under d), incrementing the current threshold value by the predetermined increment or
  • - If the fictitious decremented threshold value associated fictitious cost function is smaller than the current cost function determined under d), decrementing the current threshold value by the predetermined increment or
  • - otherwise maintaining the current threshold value,
• h) put back on step d).

The handover number can also serve as a secondary criterion. In this case, the threshold value is optimized according to the following process steps:

• a) specification of an initial threshold value,
• b) Determine the channel blocking probability, the Ka channel loss probability and the number of channels change as respective functions of this threshold a predefinable evaluation period,
• c) incrementing the threshold value by a predeterminable Increment
• d) determining the channel blocking probability, the Ka channel loss probability and the number of channels change as the respective function of this current threshold values over the predefinable evaluation period and determination an associated current cost function based on the telten values,
• e) Determination of the local slope from the canal Blocking probability function, from the channel Loss probability function and the number of Channel change based on the determined values and on it building estimation of the respective function values for the Ka channel blocking probability, the channel loss probability probability and the number of channel changes for one fictitious, incremented by the specified step size Threshold on the one hand and for a fictitious one, around the given level increment decremented threshold to the other,
• f) Determining the incremental and fictitious of the fictitious decremented threshold associated fictitious cost functions  based on the estimated values,
• G)
  • - if either
    the fictitious cost function associated with the fictitious incremented threshold value is smaller than the current cost function determined under d) or else
    the cost functions are the same and the fictitious number of channels associated with the fictitious incremented threshold value is less than the current number of channel changes determined under d), incrementing the current threshold value by the predetermined step size
    or
  • - if either
    the fictitious cost function belonging to the fictitious decremented threshold is smaller than the current cost function determined under d) or else
    the cost functions are the same and the fictitious number of channels associated with the fictitious decremented threshold value is smaller than the current number of channel changes determined under d), decrementing the current threshold value by the predetermined step size
    or
  • - otherwise maintaining the current threshold value,
• h) put back on step d).

It turned out to be favorable if the value Zero serves as the initial threshold.

According to a further advantageous embodiment of the procedure rens of the present invention, a respective cost function from the sum of the respective channel Blocking probability and the respective with a Predeterminable weighting factor multiplied channel Loss probability determined.

If a local optimum for the Threshold was found will be advantageous  Shaping the threshold always around the front given increment incremented if this threshold over a predefinable multiple of the evaluation period has remained changed. This ensures that also in the network value continues to vary, the threshold value is optimized.

It has proven to be beneficial if everyone Connection request between a mobile station and the base station or a channel from the with the current threshold determined amount of avail other channels.

Particularly good results can be achieved if the first available channel from the specified quantity of the available channels.

Other advantages and details of the present invention ben use the following illustration of an advantageous Embodiment and in connection with the figures. It shows:

Fig. 1 shows the course of the handover number H _t and the degree of service GoS in a scenario with 7 cells with 2400 fixed subscribers per km ² assuming sector antennas and subscriber assignment to the best base station, with the values on the left without , can be seen on the right-hand side with optimization of the channel allocation according to the invention, and

Fig. 2 _t the course of the handover number H and the degree of Service GoS over 35 base station locations in egg nem scenario with mobile CTM users with the process of this invention.

The following data are essentially required to carry out the method according to the invention:

• - blocking probability,
• - probability of loss and
• - Number of handovers.

This data is already collected in today's networks, e.g. B. from the OMC (Operation and Maintenance Center) for GSM networks (Global System for Mobile Communication).

The following abbreviations are used for the following detailed description of the method according to the invention:
V _t : Loss probability with threshold t
B _t : blocking probability with threshold t
COST _t : Cost function COST _t = B _t + weight.V _t , with x <0; in the case of GoS (Grade of Service): weight = 10
weight: weighting factor of the cost function COST _t
H _t : Handover requests (e.g. overall or also per connection) at the threshold t
Observ: Duration of the observation period (e.g. 1 hour)
t (i): threshold after the i-th observation period
Delta: Step size with which the threshold value is changed (e.g. 2 dB)

With each call request or handover request, each DCA procedure searches a channel for a specific algorithm. Known algorithms include:

• - first available channel (FA)
• - random channel allocation (random channel assignment, RCA)
• - Channel with the least interference on the side of the base station (least interfered channel at BS position)
• - Channel with the least interference on the user side (least interfered channel at user position)
• - geometric DCA.

Only a subset of all channels can be used, namely the channels that meet the C / I criterion (that is the signal-to-noise ratio). For example, the C / I criterion read that the received signal z. B. 12 dB more received must be considered as the sum of the noise signals.

The idea of the threshold in this case means to demand that the signal-to-noise ratio must be 12 dB + threshold. For a threshold of 6 dB would result in 18 dB. Thereby the amount of possible channels is restricted, resulting in egg lead to a higher blocking probability. The The consequence, however, is that such channels have better radio quality and therefore loss of connection and handover are untrue become more apparent.

In this sense, the Ver proposed by the invention drive to complement conventional DCA processes stand.

Central approach is that the allocated channel should be better should, as the C / I criterion plus the threshold. which DCA method is used as a basis important because the benefits can fluctuate. In principle, however an improvement is achieved for each process.

The process according to the invention is described below in the three central steps A.) to C.):
A.) The method starts with an initial threshold t (0), e.g. B. t (0) = 0. This corresponds to the original DCA algorithm.

For the observation period Observ, the data for loss probability V _t (0), blocking probability B _t (0) and the number of channel changes or handovers H _t (0) are collected.

Then will

t (1) = t (0) + delta

set.

After a second period of time are observ

V _{t (1)} = V _{t (0) + delta}

B _{t (1)} = B _{t (0) + delta}

and

H _{t (1)} = H _{t (0) + delta}

known.

The cost function COST _{t (1) can} be determined exactly from the first two values according to:

COST _{t (1)} = B _{t (1)} + weight.V _{t (1)}

If you interpret the handover rate, blocking and loss probability as functions that depend on the threshold value, then the function values at the two positions t (0) and t (1) = t (0) + delta are known. The local slope of these functions can be determined from this as follows:

V ' _{t (1)} ≈ (V _{t (1)} - V _{t (0)} ) / delta

B ' _{t (1)} ≈ (B _{t (1)} - B _{t (0)} ) / delta

H ' _{t (1)} ≈ (H _{t (1)} - H _{t (0)} ) / delta

With these values, the function values for t (1) + delta and t (1) - delta can be estimated according to:

V _{t (1) + Delta} ≈ V _{t (1)} + Delta.V ' _{t (1)}

V _{t (1) -Delta} ≈ V _{t (1)} - Delta.V ' _{t (1)}

B _{t (1) + Delta} ≈ B _{t (1)} + Delta.B ' _{t (1)}

B _{t (1) -Delta} ≈ B _{t (1)} - Delta.B ' _{t (1)}

H _{t (1) + delta} ≈ H _{t (1)} + delta.H ' _{t (1)}

H _{t (1) -Delta} ≈ H _{t (1)} - Delta.H ' _{t (1)}

For the associated cost functions, this results in:

COST _{t (1) + delta} ≈ B _{t (1) + delta} + weight.V _{t (1) + delta}

COST _{t (1) -Delta} ≈ B _{t (1) -Delta} + weight.V _{t (1) -Delta}

At this point it should be noted that the function values for t (1) - Delta are already known in this case because t (1) - Delta = t (0) applies. But in order for the following steps all To be mean enough, the calculation is explicit here give.

The value of t (2) finally results from this as follows:

t (2) = t (1) + delta, if

COST _{t (1) + Delta} <COST _{t (1)} or

COST _{t (1) + Delta} = COST _{t (1)} and H _{t (1) + Delta} <H _{t (1)} .

t (2) = t (1) - delta, if

COST _{t (1) -Delta} <COST _{t (1)} or

COST _{t (1) delta} = COST _{t (1)} and H _{t (1) delta} <H _{t (1)} .

t (2) = t (1), in all other cases (in this case the Threshold already optimized locally).  

B.) The step from t (i) to t (i + 1) is then analogous to Step from t (1) to t (2) shown above for all i <1.

If the local optimum is found, then the method according to the invention no longer changes the found threshold value t _opt . Only data for the value t _{opt are} collected, with which the cost function COST _topt and the number of handovers H _{topt can} be calculated exactly.

However, the previously determined values COST _{topt + Delta} and H _{topt + Delta} or COST _topt-Delta and H _topt-Delta can quickly become obsolete. In order to continue to find the optimal threshold value even in varying network situations, these values are updated from time to time according to the invention.

This can be done using the following rule:
C.) If z. B. For 10 consecutive observation periods the threshold value has not changed, then the threshold value is increased by delta.

If the local optimum has not changed, then the threshold value according to the Ver move around Delta again in the next step Ringert.

It is important for the implementation of the method according to the invention that the data collected is statistically representative, ie the blocking probability determined should, for. B. have not been determined on the basis of two or three blockages, but, for example, on the basis of a larger number, such as 100. This can be controlled by the length of the observation period Observ according to:

• - the larger ob, the better the statistical data
• - but the smaller the observation, the faster the process can react to changes in the network

The choice of delta step size also has a major influence on the result:

• - The larger delta, the faster the process can open React to changes in the network
• - The smaller delta, the more precise the optimum can be be approximated

When using the invention, there are several options ner application to different parts of a radio network, näm on the entire network, on a subnet or even on individual base stations. In practice, this also depends the system architecture. For the example of a GSM network For example, all base stations connected to an OMC can use the use the same optimized threshold.

In the following simulation examples, the results are shown in the Figures Fig. 1 and Fig. 2, a homo genes network 7 (Fig. 1) and 35 (Fig. 2) locations base station and a first-available-DCA algorithm based on.

The method according to the invention, however, takes place for each scene nario the threshold with the optimal performance if it there is such an optimum. If there are several optima, will found a local optimum.

If there are several optima, this can also be a sign of to be bad statistical data. In this case it should the observation period can be chosen longer.

In the case of cellular mobile radio systems, a distinction is made, among other things, between the following different handover methods, in which mobile networks are judged according to the GSM standard, which administrative areas of the network are changed and which parts of the network infrastructure will be affected. One differentiates:

• - Intracell handover: handover within a cell without cell change, but with change of radio frequency.
• - Intercell handover: handover between neighboring Cells, but from the same Base Station Controler (BSC) are managed.
• - External Handover: Handover between two cells managed by different base station controllers become.
• - Basic handover: handover between two cells managed by different base station controllers be in turn at different, with each other connected Mobile Switching Center (MSC).
• - Subsequent handover: handover between two cells, managed by different base station controllers be, in turn, at different, however MSC not connected so that a third MSC must be switched on. That guy Handover is usually based on a basic Handover when the MS is in the area of influence third MSC moves.

The inventor used extensive simulations to investigate the influence of the threshold value in various scenarios. The following threshold values were simulated:
0, 2, 4, 6, 8, 10, 15, 20 and 30 dB. The usual grade of service GoS was assumed for the cost function COST, ie weight = 10.

Scenario 1 ( Fig. 1)

7 cell scenario with 2400 fixed subscribers per km ² assuming sector antennas and subscriber assignment to the best base station. The optimal threshold is 6 dB.

The GoS has halved for the marginal cells from 1 to about 0.5. The number of handovers H _{t is} reduced by the method from approx. 7000 to only 2000. This relationship is recorded in the illustration according to FIG. 1, in which the GoS and the Intracell handover requests H _t are plotted in such a way that on the left The values without can be seen on the right side with optimization of the channel allocation according to the invention. From this, the gain through the inventive method can be seen impressively.

Scenario 2

7 cell scenario with 400 fixed participants per km ² . Omni antennas were used as the basis for assigning participants to the best base station. The optimal threshold is 20 dB. With a GoS of 0, the number of handovers decreases from 6000 or 9000 to 120 or 180.

Scenario 3

7 cell scenario with 1200 fixed participants per km ² . Omni antennas were used as the basis for assigning participants to the nearest base station. The optimal threshold is 0 dB, since the GoS is minimal here. So the handovers don't matter.

Scenario 4

19 cell scenario with 400 fixed participants per km ² . Omni antennas were used as the basis for assigning participants to the best base station. The optimal threshold is 8 dB. With a GoS of 0, the number of handovers for the center cell decreases from 9500 to approximately 600.

Scenario 5 ( Fig. 2)

There are 35 base station locations and mobile CTM participants. The optimal threshold is 8 dB. The GoS is almost halved (compared to a threshold of 0). The Intracell Handover Requests Ht decrease from over 450,000 to less than 150,000, while the Intercell Handover Requests are only slightly reduced. The different behavior with Intracell and Intercell Handover Requests is easy to understand, because Intercell Handover Requests arise when a participant physically changes from one cell to another cell due to his mobility. As a result, he usually forces an intercell handover. Therefore, Inter cell handover requests can only be reduced slightly. This scenario is recorded in the illustration according to FIG. 2, in which the GoS and the number of Intracell handover requests H _t per hour are plotted over the different base station locations.

With the present invention, the following advantages result in wesent union:

• - The performance of a radio network (blocking and Loss probability and handover number) is improved
• - The process can be based on any DCA process to optimize it,
• - It is a generalizable method for Threshold optimization, which can also be used to vary others Threshold values such as B. used for handover threshold values who can
• - The method according to the invention adjusts the threshold respective network, as well as changes in reacts within a network (e.g. when changing of traffic or if a base station fails Etc.),
• - The procedure does not lead to deterioration in any case grid performance because if the conventional DCA Process was optimal in this sense, then it will methods according to the invention select the threshold value equal to 0. The corresponds to the original DCA procedure without threshold. This statement is based on the assumption that the statistical Data is good enough, and ignores that every now and then for for a short time the optimal threshold value is increased by delta (see step C).

Claims (9)

1. Method for optimized channel allocation between a base station (BS) and a plurality of mobile stations (MS) in a radio system, in particular in a cellularly constructed mobile radio system, with a respective signal-to-noise ratio (C / I) of a received signal for possible interference nals serves as a selection criterion for a number of channels for channel allocation, characterized in that this respective signal-to-noise ratio (C / I) is increased by a predetermined threshold value (t) and the selection of channels for channel allocation to the quantity of such available channels that meet this extended selection criterion is restricted, the threshold value being adaptively varied in predefinable time intervals such that a channel blocking probability (B _t ) and / or channel loss probability (V _t ) become optimal. 2. The method for optimized channel allocation according to claim 1, characterized in that the optimization of the threshold value (t) takes place according to the following procedural steps:

• a) specification of an initial threshold value (t (0)),
• b) determining the channel blocking probability (B _t ) and the channel loss probability (V _t ) as the respective function of this threshold value (t) over a predefinable evaluation period (ob),
• c) incrementing the threshold value (t) by a predeterminable step size (delta),
• d) Determining the channel blocking probability (B _t ) and the channel loss probability (V _t ) as the respective function of this current threshold value over the predefinable evaluation period (Observ) and determining an associated current cost function (COST _t ) on the basis of the determined values .
• e) Determination of the local slope of the channel blocking probability function (B ' _t ) and of the channel loss probability function (V' _t ) on the basis of the values determined and the subsequent estimation of the respective function values for channel blocking probability (B _{t + delta} , B _t-delta ) and channel loss probability (V _{t + delta} , V _t-delta ) for a fictitious threshold value (t + delta) incremented by the predefined increment on the one hand and for a fictitious threshold value by the predefined increment decremented threshold (t - delta) to the other,
• f) determining the fictitious cost functions associated with the fictitious incremented and fictitious decremented threshold value (COST _{t + delta} , COST _t-delta ) on the basis of the estimated values,
• g) provided that t the fictitious incremented threshold value (+ delta) associated notional cost function (COST _{t + delta)} klei ner than the current under d) determined cost function (COST _t), incrementing the current threshold value by the predetermined step size (Delta) or
  if the fictitious decremented threshold (t-delta) associated fictitious cost function (COST _t-delta ) is smaller than the current cost function (COST _t ) determined under d), decrement the current threshold by the specified step size (delta) or
  otherwise maintaining the current threshold value,
• h) put back on step d).

3. The method for optimized channel allocation according to claim 1, characterized in that the optimization of the threshold value (t) takes place according to the following procedural steps:

• a) specification of an initial threshold value (t (0)),
• b) determining the channel blocking probability (B _t ), the channel loss probability (V _t ) and the number of channel changes (H _t ) as respective functions of this threshold value (t) over a predefinable evaluation period (Ob),
• c) incrementing the threshold value (t) by a predeterminable step size (delta),
• d) Determining the channel blocking probability (B _t ), the channel loss probability (V _t ) and the number of channel changes (H _t ) as a function of this current threshold value (t) over the predefinable evaluation period (Ob) and determining an associated current one Cost function (COST _t ) based on the determined values,
• e) Determination of the local slope of the channel blocking probability function (B ' _t ), of the channel loss probability function (V' _t ) and of the number of channel changes (H ' _t ) on the basis of the determined values and the estimation based thereon the respective function values for the channel blocking probability (B _{t + delta} , B _t-delta ), the channel loss probability (V _{t + delta} , V _t-delta ) and the number of channel changes (H _{t + delta} , H _{t- Delta} ) for a fictitious threshold value (t + delta) incremented by the specified step size and for a fictitious threshold value (t - delta) decremented by the specified step size on the other
• f) determining the fictitious cost functions (COST _{t + delta} , COST _t-delta ) associated with the fictitious incremental and fictitious decremented threshold value on the basis of the estimated values,
• g) if either
  the fictional cost function (COST _{t + delta} ) associated with the fictitious incremental threshold value is smaller than the current cost function (COST _t ) determined under d) or the cost functions (COST _t , COST _{t + delta} ) are the same and those that are incremental to the fictitious one Fictitious number of channel changes (H _{t + Delta} ) associated with the threshold value is smaller than the current number of channel changes (H _t ) determined under d), incrementing the current threshold value by the predetermined step size
  or
  provided either
  the fictional cost function (COST _t-delta ) associated with the fictitious decremented threshold value is smaller than the current cost function (COST _t ) determined under d) or the cost functions (COST _t , COST _t-delta ) are the same and the fictitious decremented threshold value associated fictitious number of channel changes (H _t-Delta ) is smaller than the current number of channel changes (H _t ) determined under d), decrementing the current threshold value by the predetermined step size
  or
  otherwise maintaining the current threshold value,
• h) put back on step d).

4. A method for optimized channel allocation according to claim 2 Or 3, characterized in that the value zero serves as the initial threshold value (t (0)). 5. The method for optimized channel allocation according to one of claims 2 to 4, characterized in that a respective cost function (COST _t ) from the sum of the respective channel blocking probability (B _t ) and the respective channel multiplied by a predeterminable weighting factor (weight) -Loss probability (V _t ) is determined. 6. Procedure for optimized channel allocation according to one of the Claims 2 to 5, characterized in that the threshold value (t) always again by the predetermined step width (delta) is incremented if this threshold value ü over a predefinable multiple of the evaluation period (loading whether) has remained unchanged. 7. Procedure for optimized channel allocation according to one of the Claims 2 to 6, characterized in that for every connection request between a mobile station (MS)  and the base station (BS) a channel from the one with the current one Threshold (t) certain amount of available channels is shared. 8. Procedure for optimized channel allocation according to one of the Claims 2 to 7, characterized in that with every channel change request a channel from the one with the current threshold (t) certain amount of available channels is allocated. 9. Procedure for optimized channel allocation according to one of the Claims 7 or 8, characterized in that the first available channel from the specified quantity of the available channels.