Method and System for Mitigating Inter-Cell Interference

TECHNICAL FIELD

The present invention relates in general to a method and a system for mitigating inter-cell interference and in particular to a method and a system for transmission power determination and allocation to mitigate inter-cell interference.

BACKGROUND

Orthogonal Frequency Division Multiplexing (OFDM) is a promising transmission technique for broadband wireless networks. It has already been adopted by, for instance, Digital Video Broadband-Terrestrial (DVB-T) and Asymmetrical Digital Subscriber Line (ADSL) and was recently proposed as a radio transmission scheme for 3G-Long term Evolution (LTE) cellular systems. An important feature of OFDM is the orthogonality among the subcarriers, which reduces or eliminates intra-cell interference. Nevertheless, inter-cell interference exists and may degrade the overall performance of the network. Specifically, in the downlink direction the impact of inter-cell interference may be more pronounced for cell-edge users, due to the relatively low channel gains to their respective serving base stations and due to the proximity of neighbour base stations.

The problem of resource allocation in the downlink direction of multi-cell OFDM networks has been extensively studied. Early studies on Radio Resource Management (RRM) by Su and Geraniotis, 1998 (H. J. Su and E. Geraniotis, “A Distributed Power Allocation Algorithm with Adaptive Modulation for Multi-Cell OFDM Systems,” in IEEE International Symposium on Spread Spectrum Techniques and Applications, 1998) and Kiani et al., 2006 (S. Kiani, D. Gesbert, J. E. Kirkebø, A. Gjendemsjø and G. E. Øien, “A Simple Greedy Scheme for Multicell Capacity Maximization,” in International Telecommunications Symposium, 2006) reduced the complexity of the resource allocation problem by assuming a single user per cell. In these studies the subcarrier assignment problem becomes effectively equivalent with the power allocation problem since the user has access to the whole bandwidth and thereby the complexity of the problem is greatly reduced.

These studies follow similar processes in the sense that they formulate a total rate maximization problem subject to power constraints. In order to solve this problem they propose distributed offline heuristic algorithms, which perform power allocation based on the well known water-filling algorithm.

The problem of power allocation in a multi-cell environment is a non-convex problem that has not been solved optimally in a distributed way in polynomial time.

Moreover the convergence speed of the distributed algorithms in Su and Geranoitis, 1998; and Kiani et al., 2006 is relatively low. The convergence of such algorithms would be even more challenging in systems like LTE, which primarily rely on fast packet scheduling and access to the total available spectrum is facilitated with a fine granularity through the dynamic allocation of frequency channels.

In Li and Liu, (G. Li and H. Liu, “Downlink radio resource allocation for multi-cell OFDMA system” in IEEE Transactions on wireless communications, vol. 5, No. 12, December 2006), a radio resource allocation system is presented wherein the power to be allocated is determined as a flat power being the total transmission power of the base divided by the total number of resource blocks. Herein, a power is allocated without taking into account inter-cell interference and the focus is on determining the assignment of frequency channels to users.

There is hence a need for a method and a system of power and channel allocation that diminishes and/or eludes at least some of the problems as mentioned above.

It should be emphasized that the term “comprises/comprising” when being used in the specification is taken to specify the presence of the stated features, integers, steps or components but does not preclude the presence or addition of one or more other features, integers, steps or components or groups thereof.

SUMMARY

An object of the present invention may be to provide near optimized channel and transmission power allocation mitigating inter-cell interference.

According to one aspect there is provided a method for determining a transmission power level for transmission power allocation in a communication system, comprising the steps of:

- obtaining channel gain data related to at least a first user equipment and its serving first base station,
- obtaining performance related data related to at least the first user equipment,
- adapting of a transmission power level per group of resource blocks for at least the serving first base station for transmission related to the at least first user equipment, for increasing a calculated data bit throughput of the communication system, dependent on the obtained channel gain data and the obtained performance related data, and
- providing the transmission power level per group of resource blocks for power allocation as an upper limit of the at least serving first base station transmission power on the super frame level,
  
  wherein the super frame level comprises a level for at least one frame, and wherein the group of resource blocks comprises at least one resource block.

Said step of adapting of a transmission power level may further comprise reallocating power that is associated with resource blocks that are not used by the serving first base station at the time of power determination to at least one group of resource blocks that is used by the serving first base station at the time of power determination, in the case the calculated throughput of system increases with an increased allocated power.

The performance related data in the step of obtaining performance related data may comprise one or more of the following: buffer status at the base station (BS) for the at least first user equipment, type of services related to the at least first user equipment at the time of power determination, throughput data for the at least first user equipment during a past super frame, channel quality indicator (CQI) input data, and channel state information (CSI) input data.

According to another aspect, there is provided a method for processing of transmission power level related data in a base station, related to a communications system, comprising the steps of:

- obtaining channel gain data from at least a served first user equipment (UE),
- providing performance related data for at least the served first user equipment,
- obtaining determined transmission power level data per group of resource blocks, as determined at a super frame level,
- providing transmission power related pilot signals to at least the served first user equipment,
- obtaining channel quality related information from at least the served first user equipment, and
- adapting of the transmission power level per resource block on the frame level, comprising altering the determined transmission power level as determined per group of resource blocks at the super frame level, in dependence of the obtained channel quality related information related to at least the served first user equipment and the obtained channel gain data.

The channel quality related information in the step of obtaining channel quality related information may further comprise a signal to interference and noise ratio target.

The step of altering the determined transmission power level per resource block on the frame level may further comprise decreasing the transmission power level as determined on the super frame level, in dependence of the obtained signal to interference and noise ratio target and the obtained channel gain data, related to at least the served first user equipment.

The step of altering the determined transmission power level per resource block on the frame level may further comprise increasing the transmission power level on a frame level from the level as determined on a super frame level, in dependence of the obtained signal to interference and noise ratio target, whereas maintaining the average transmission power level over all resource blocks used by the served base station equal or below the average transmission power level over all resource blocks as determined on the super frame level.

The channel gain data in the step of obtaining channel gain data may comprise average data over fast fading channel gains.

The method for processing transmission power allocation related data may further comprise calculating average data over fast fading channel gains from the obtained channel gain data.

The channel quality information in the step of obtaining channel quality information, may comprise substantially instantaneous channel quality related information.

The method for processing transmission power allocation related data may further comprise providing pilot signals to at least the served first user equipment.

According to yet another aspect, there is provided a user equipment for providing channel gain data by a user equipment related to transmission power allocation in a communications system, comprising the steps of:

- providing channel gain data for the base station serving said user equipment,
- obtaining pilot signal information, comprising a transmission power level related information as determined per group of resource blocks on a super frame level, and
- providing channel quality related information, based on the obtained pilot signal information.

The channel gain data in the step of providing channel gain data may further comprise average data over fast fading channel gains.

The channel quality related information in the step of providing channel quality related information may comprise substantially instantaneous channel quality related information.

The method for providing channel gain data may further comprise obtaining regular pilot signal information, and wherein providing channel gain data for a base station may be based on the obtained first pilot signal information.

According to yet another aspect, there is provided a radio network controller, arranged to obtain channel gain data related to at least a first user equipment and its serving first base station, obtain performance related data related to at least the first user equipment, adapt a transmission power level per group of resource blocks for at least the serving first base station for transmission related to the at least first user equipment, for increasing a calculated data bit throughput of the communication system, dependent on the obtained channel gain data and the obtained performance related data, and provide the transmission power level per group of resource blocks for power allocation as an upper limit of the at least serving first base station transmission power on the super frame level, wherein the super frame level comprises a level for at least one frame, and wherein the group of resource blocks comprises at least one resource block.

The radio network controller may further comprise a regulator unit that is arranged to determine a number of resource blocks per user equipment, and may further comprise a power determination unit arranged to determine a long-term power allocation level on a super-frame level per group or resource blocks, wherein the group of resource blocks comprises at least one resource block and the super frame comprises at least one frame.

According to yet another aspect, there is provided a base station unit for a radio communication system, arranged to obtain channel gain data from at least a served first user equipment (UE), provide performance related data for at least the served first user equipment, obtain determined transmission power level data per group of resource blocks, as determined at a super frame level, provide transmission power related pilot signals to at least the served first user equipment, obtain channel quality related information from at least the served first user equipment, and adapt the transmission power level per resource block on the frame level, comprising altering the determined transmission power level as determined per group of resource blocks at the super frame level, in dependence of the obtained channel quality related information related to at least the served first user equipment and the obtained channel gain data.

The base station unit may further comprise a gain control unit arranged to provide gain data related to at least a first user equipment, and may comprise a power allocation unit arranged to allocate transmission power for communication with at least the first user equipment, based on a transmission power level.

The transmission power level within the base station may further comprise at least a long-term transmission power level determined on a super-frame level.

The transmission power level within the base station may further comprise a short-term transmission power level determined on a frame level.

According to yet still another aspect, there is provided a user equipment for a radio communication system, arranged to provide channel gain data for the base station serving said user equipment, obtain pilot signal information, comprising a transmission power level related information as determined per group of resource blocks on a super frame level, and provide channel quality related information, based on the obtained pilot signal information.

The user equipment may further comprise a gain unit arranged to provide gain data to the base station serving the user equipment, and may comprise a radio channel unit arranged to provide channel quality related data to said base station.

The user equipment may further comprise a mobile phone.

The term resource block (RB) used in the present invention stands for a physical time-frequency resource in OFDM access (OFDMA). Typically more than one resource block is allocated for data transmission. In evolved universal Terrestrial Radio Access Network (E-UTRAN) one resource block comprises 12 sub-carriers each having 15 kHz carrier spacing over 0.5 ms, i.e. 1 RB=180 kHz×0.5 ms.

In OFDMA as well as in E-UTRAN each frame is typically 10 ms long in time.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to explain advantages and features herein in more detail a few embodiments will be described below, where references are made to the accompanying drawings, for which

FIG. 1 illustrates a message sequence diagram according to the prior art;

FIG. 2 presents method steps illustrating features of power allocation according to some embodiments;

FIGS. 3-5 illustrate various aspects of resource allocation;

FIG. 6 illustrates a communication system according to some embodiments; and

FIG. 7 illustrates a message sequence diagram according to at least some embodiments.

DETAILED DESCRIPTION

The algorithms of the prior art documents, as mentioned above, are localized and hence a new estimation of the interference is needed every time a BS tries to allocate power to its own subcarriers. Consequently, this may lead to an increase in the number of pilot signals transmissions.

Moreover, a fully centralized algorithm which coordinates power allocation over the whole cellular system is not feasible from a computational, delay and signalling overhead perspectives, according to Koutsopoulos and Tassiulas; 2006.

In addition, the existing prior art does not take into account explicit, by way of minimum, peak or average, rate requirements for admitted sessions.

FIG. 1 shows a message sequence diagram of a prior art system. At the beginning of each frame, the BSs may transmit pilot signals step 102, which can be used by the UEs in order to estimate an instantaneous channel profile, for example channel quality indicator (CQI) or channel state information (CSI). These measurements can be reported back, step 104, to the serving BS, which performs scheduling and power allocation without considering inter-cell interference. Each BS may then transmit data along with pilot signals to its served users, step 106. These new pilot signals can now be used for the estimation of the channel gains of the next frame, step 108, which could then be used for further downlink scheduling and power allocation, step 110.

Signalling is thus taken place between the UEs and the serving BSs only, as well as on a single timescale.

The underlying idea of at least some embodiments according to the present invention is to allocate power using a dynamic scheme, where power allocation may occur in two timescales.

At the first level, having a slower time scale, in the order of seconds, power allocation may be performed using a network-wide knowledge of the long-term channel gains, or long-term radio conditions of the users, and the long term user performance criteria and status, for example achieved user throughput, buffer size, bit rate requirement etc. The width of this level, which will be referred to as the super-frame level, is to handle inter-cell interference and to obtain a close to optimum network-wide power allocation.

The second level, which is performed at a faster time scale, in the order of milliseconds, power allocation may be performed locally at each BS using the constraints on power levels imposed by the super-frame level as well as on the instantaneous channel gains or short-term radio conditions of the users. The width of this level, which will be referred to as the frame level, is to exploit multipath propagation and fast fading.

In addition, the power allocation scheme according to some embodiments may take into account rate requirements, corresponding to a number of resource blocks RB per user, assuring fair allocation, with respect to their defined rate requirements, of the resources among users.

In the following a detailed example according to at least some embodiments is presented.

Considering an Orthogonal Frequency Division Multiplexing Access (OFDMA) system with L Base Stations (BSs), which are deployed in a hexagonal grid and each BS is positioned in the middle of each cell. Let L denote this set of BSs and M_lthe set of users connected to BS l. The number of users in the l^thBS is thus M_land the entire network has in total

$M = \sum_{l = 1}^{L} M_{l}$

users.

Radio resources may be partitioned in both the frequency domain and the time domain. The total available bandwidth may be divided into N sub-bands, where each sub-band is a cluster of k consecutive OFDM subcarriers.

In addition, the time resource may be divided into frames, while a number of consecutive frames construct a super-frame. The smallest resource unit that can transport data is a combination of one frequency resource unit (sub-band) and one time resource unit (frame). This basic resource unit corresponds to one Resource Block (RB).

According to at least some embodiments, the Radio Resource Control (RRC) that is responsible for resource allocation may be realized at both a Radio Network Controller (RNC) and base stations (BSs). This semi-distributed RRC scheme may reduce the computational load by splitting the decisions in two levels, a RNC level and a BS-level.

The RNC algorithm may be executed at a slower time scale and may allocate resources on a super-frame level based on the long-term channel profile as well as on the long-term achieved performance of each user. The width of the RNC entity may be to control the interference among its BSs and maximize the interference avoidance gain.

The BS algorithm, however, may be executed at a faster time scale and may allocate resources on a frame level based on the instantaneous channel profile of each user as well as on the decisions and/or restrictions of the RNC algorithm.

According to some embodiments the RNC refers to a centralized logical node. According to some further embodiments the RNC can also be realized in a distributed system provided the base stations can communicate with each other. In such systems, one of the base stations could be configured to act as a centralized server, which would collect the necessary super-frame information from the base stations and re-distribute it to the desired set of base stations for using it for localized resource allocation.

In the following a description of system operation during one super-frame cycle, is presented. At this time scale, corresponding to one super-frame, long-term power allocation may be performed, which in turn may be used as an input to individual base stations for their respective distributed power allocation. This is described further below.

Centralize or RNC Based Iterative Power and Channel Allocation: the Longer Time Scale

One super-frame corresponds to the acting period of the RNC decisions and it typically lasts several scheduling instances, being referred to as transmission time intervals, (TTIs). Each BS transmits reference or pilot signals for enabling the User Equipment (UE) to perform downlink measurements and measurement reporting, facilitating channel estimations. Specifically, based on the received signals each UE may calculate the long-term channel gains or any relevant long-term downlink measurement, which may average out fast fading. Each UE then reports these measurements to the RNC if it exists as a physical node. Otherwise in a distributed architecture, each UE reports these measurements to the serving BS, which distributes this information to the other BSs of the cluster, for example through the standardized X2 interface in LTE.

By using these measurements the RNC, either represented by a logical node or a physical node, may construct the average gain matrix G of the network, where UEs lie in the first dimension and BSs in the second one. Here, the dimension of G matrix is M-by-L. Based on matrix G, the Shannon limit capacity of a RB n when used by a user i can be calculated form the following formula:

$T_{in} (P) = B_{RB} \cdot \log_{2} (1 + \frac{P_{\hat{i} n} \cdot G_{i \hat{i}}}{σ_{RB}^{2} + \sum_{\underset{j \neq \hat{i}}{j = 1}} P_{jn} \cdot G_{ij}})$

wherein B is the bandwidth of a RB, G_ijis the long-term gain between BS j and user i, G_iîis the long term gain between user i and its serving BS î, P_în, is the transmission power of the useful signal at RB n, P_jn, is the transmission power of the interfering signal from the j^thbase station at RB n and σ_RB²is the variance of the additive white Gaussian noise (AWGN) with zero mean value.

The objective of the RNC may thus be to maximize the overall throughput of the system by allocating power to RBs. This is termed centralized power allocation or allocation by RNC, being either logical or physical. Specifically, the optimization problem can be formulated as follows:

$\begin{matrix} \underset{P}{Max} \sum_{i = 1}^{M} \sum_{n = 1}^{N} T_{in} (P) & (P .1) \end{matrix}$

subject to:

$\begin{matrix} \sum_{n = 1}^{N} \sum_{m \in M_{l}} P_{mn} \leq P_{tot}, l = 1, \dots, L & (C .1 .1) \\ P_{mn} \geq 0, m = 1, \dots, M; n = 1, \dots, N & (C .1 .2) \\ R_{i, \min} \leq \sum_{n = 1}^{N} T_{in} \leq R_{i, \max}, i = 1, \dots, M & (C .1 .3) \end{matrix}$

Algorithm Description

The optimization problem characterized by (P.1) is in general difficult to be solved due to its non-convex characteristics. In order to solve it, it is proposed an iterative heuristic algorithm. The algorithm comprises two parts:

In the first part, the algorithm may allocate RBs to the users based on users' current bit rate requirement and on the assumption of equal power distribution among RBs. The first parameter, the users' current bit rate/performance requirement may in turn be derived at the RNC by two main sets of parameters, user service type/requirement and its achieved performance (i.e. in the last super-frame). This is further elaborated down below.

Since different users could have different performance, the first iterative step may not necessarily allocate the same number of RBs to all users. In practical situation it is also likely that some of the RBs remain unused in at least some base stations. This property is exploited in the second iterative step.

In the second part, the algorithm attempts to maximize further the system throughput by allocating any excessive power, which is residual from the first part due to unused RBs, to already assigned RBs.

With reference to FIG. 2, presenting a flow-chart overview of steps of a radio network control related allocation algorithm is described. The method may start with selecting a first BS in step 202, here BS k. Thereafter, the throughput contribution to the system of each UE m when being allocated RB n, is calculated for all UEs. In the following step, step 206 it is determined whether the k is the last BS, that is BS L, or not. If it is determined that the index k does not equal index L, another BS may be selected with incremented index, in step 202. If it is on the other hand determined that k equals L in step 206, the following step is selecting a BS k and calculate the residual power ΔP, in step 208. The residual power is the power that is associated with a resource block that is not used by the current BS k. Therefore, this power may be used elsewhere.

The next step is selecting a RB from the set of Nk, that is the set of RBs that are used by the kth BS, and allocate additional power ΔP. A determination as to whether the throughput is increased or not, with additional power, is then performed in step 212.

In the case the throughput is not increased in step 212, the next is step 214, updating the residual power in step 214. If it is determined in step 212 that the throughput is increased the following is step 216, determining whether hare are any available RBs for BS, or not, in step 216. If it is determined that are more RBs available in step 216, the next step is step 210, selecting a RB from the set of Nk etc. If it is determined in step 216 that there are no available RBs for BS it is determined whether k equals L or not. In the case k does not equal L, the following step is to select a BS k and to calculate residual power, in step 208.

However, if it is determined that k equals L in step 218, the method is ended in step 220. Residual power may thus be allocated to RBs that are used by the BS k.

In the following, the pseudo code of the RNC allocation algorithm is given.

TABLE 1

Pseudo code variables of the RNC allocation algorithm.

Variable
Meaning

L
Number of BSs controlled by the RNC

M
Total number of users in the system

N
Total number of available RBs

M_k
Number of users connected to the k^thBS

T_bef(m, n)
Total throughput of the system before

allocating RB n to user m

T_aft(m, n)
Total throughput of the system after

allocating RB n to user m

Ω(m, n)
Throughput contribution to the system

of user m when is allocated RB n

X
L-by-N index matrix which defines the

RB allocation among BSs

Y
M-by-N index matrix which defines

the RB allocation among users

P
M-by-N long-term power allocation

matrix

e_k^L
A vector with length L where its

elements are all zero except the k^th

element

N_k
Set of RBs used by the k^thBS

Ñ_k
Set of RBs of BS k that do not take

part in the power redistribution

procedure

Initialization: X=zeros(L, N); Y=zeros(M, N);

For k = 1 : L do % Base station loop

y_Mtx1= [0 , 0 , . . . , 0]^T;

x_Lx1= [0, 0, . . . ,0]^T;

T_bef= zeros(M_k, N);

T_aft= zeros(M_k, N);

Ω = [ ];

For m ε M_kdo % User loop

For n = 1 : N do % RB loop

x = X(1 : L, n);

y = Y(1 : M_t, n);

T_bef(m,n) = T_n(y)

x = x + e_k^L;

y = y + e_m^M;

T_aft(m,n) = T_n(y);

End For

End For

Ω = Taft − Tbef ;

(X (k, 1 : N), Y (m \in M_{k}, 1 : N)) \leftarrow \arg \max_{z} \sum_{i = 1}^{M_{k}} \sum_{j = 1}^{N} z_{ij} Ω_{ij}

\begin{matrix} subject to : & 1) \sum_{i = 1}^{M_{k}} z_{ij} \leq 1, \forall j 2) n_{i . \min} \leq \sum_{j = 1}^{N} z_{ij} \leq n_{i, \max}, \forall i 3) z_{ij} \in {0, 1} \end{matrix}

End For

P = \frac{P_{tot}}{N} \cdot Y; % Initialization of the power matrix

For k = 1 : L do

If \sum_{j \in N_{k}}^{} x_{kj} < N % Power redistribution

Δ P = \frac{P_{tot} - (\frac{P_{tot}}{N} \cdot (\sum_{j \in N_{k}}^{} x_{kj}))}{\sum_{j \in N_{k}}^{} x_{kj}};

For n E Nk

IF T_n(P(k, n) = P(k, n) + ΔP) > T_n(P(k,n))

P(k, n) = P(k, n) + ΔP;

ELSE

Ñ_k= Ñ_k∪ n;

Δ P = \frac{P_{tot} - (\frac{P_{tot}}{N} \cdot (\sum_{\underset{j \notin {\tilde{N}}_{k}}{j \in N_{k}}}^{} x_{kj}))}{\sum_{\underset{j \notin {\tilde{N}}_{k}}{j \in N_{k},}}^{} x_{kj}};

End IF

End For

Else

Continue;

End IF

End For

Resource Allocation Priority Order

The impact of the order with which the BSs are evaluated may be important for the final outcome. The reason being that the first examined BSs have interference free RBs and hence have an advantage in allocating RBs. Based on this observation, this problem can be addressed by changing at each super-frame the order of BSs evaluation. The order can be random. In that case the BS resource allocation priority order is determined by their respective probabilities: [ρ_i, ρ₂, . . . , ρ_N], where

$\sum_{i = 1}^{N} ρ_{i} = 1.$

According to at least one embodiment, all BSs have the same resource allocation priority order probability, i.e.

$ρ_{i} = \frac{1}{N}; \forall i,$

According to some other embodiments, the allocation priority may be based on the load at the BS. For example heavily loaded BSs have a preference to be served first or BSs which were not able to transmit sufficient data during past super-frames are now given the priority to be evaluated first.

According to yet other embodiments, the allocation may be started at BSs in a round robin fashion.

Optimality Conditions of Algorithm

By using the solution of the above described optimization problem, RNC may allocate power to RBs and achieve a close to optimum solution. However, this solution is valid only when accounting average, rather than instantaneous, channel gains. The output of the RNC algorithm is the matrix P, which indicates the power allocation at each RB and therefore the RB assignment to each user in case of zero power allocation on a specific RB.

An example of output from the RNC power allocation over RBs for a specific BS is illustrated in FIG. 3 exemplifying centralized or RNC level power allocation. As mentioned above, the specific power allocation has a global optimum only when accounting average channel gains. In other words, if BSs follow strictly the recommendations of the RNC, then the RNC power allocation may not anymore be optimum since it does not consider the fluctuations of the channel gains due to multipath propagation. However, the knowledge of the instantaneous radio conditions may be available at the BSs, through CQI reports at the TTI level. Therefore BSs may increase further the system throughput by exploiting fast fading, while at the same time having to take into account the decisions of the RNC algorithm.

Each BS may only be allocated half of the total power or no power at all on resource blocks that their neighbouring base stations use for user equipments that are located at or near their cell edge, close to the interfering base station. This further causes a low inter-cell interference.

Radio Measurements for Centralized Allocation

As stated above the centralized power allocation may be based on two sets of measurements, one being radio related, that is long term radio conditions that should filter out the effect of fast fading. Therefore, in principle the required measurements for the centralized power allocation could also be done in the uplink directions among the UEs and the BSs.

Nevertheless, downlink measurements are preferred due to different uplink and downlink antenna configurations as well as due to support of Time Division Duplex (TDD) mode in the specifications of Long Term Evolution (LTE) systems. In addition, the UE may already report the downlink measurements based on long term averaging to the serving BS for the purpose of handovers. Thus, the measurements needed for deriving the centralized power levels for each UE can be easily obtained.

Distributed or BS Based Power Allocation: Shorter Time Scale

After solving the optimization problem as described above (P.1), RNC may distribute its decisions, that is basically the P matrix, to the cluster of BSs that it controls. These decisions are valid for the whole duration of the current super-frame and they may be updated at the beginning of the next super-frame when RNC algorithm will be again executed.

Hence, for the current super-frame, each BS knows which RBs, are allowed to use and in addition knows exactly the sets of RBs that the other BSs will use. In other words, after the execution of the RNC algorithm, the interference may be determined for the whole super-frame period. Now, the objective of each BS is to improve further the system throughput based on the RNC decisions and user's instantaneous channel profile, for example Channel Quality Indicator (CQI).

At the BS level there are two strategies that can be adopted in order to improve further the throughput of the system. These strategies are based on the maximum power limit that may be set on each resource block during each TTI of the super-frame by the RNC.

According to state of the art systems the maximum power limit is generally a fixed parameter specific to each service type, for instance maximum 1 Watt for speech users, 2 Watts for packet data rate services of type 1, 3 watts for packet data rate services of type 2 and so on.

This way of setting the max power limit according to the state of the art does not take into account the impact of interference in the neighbor cells.

Strict Power Constraint Strategy

Using this strategy, each BS is prohibited to allocate more power on a RB than the one determined by the RNC, according to at least some embodiments; see FIG. 4 showing distributed power allocation based on strict maximum power limit.

Based on the SINR target of each mobile and on the instantaneous channel gains, or CQI, the BS allocates the final transmit power which is upper limited by the power allocated by the RNC. In the case of favoured channel gain with the serving BS and/or low channel gains with the interfering BSs, the BS can allocate a smaller amount of power on the specific RB, hence producing less interference in the system.

In other words, a strict or deterministic power level is provided to the base station by the centralized power allocation node. Thus, the BS is never allowed to exceed this limit.

This is a conservative approach from the throughput perspective. However, it ensures that interference in the neighbour cells is minimized.

Flexible Power Constraint Strategy

Using this strategy, and referring to FIG. 5 showing distributed power allocation based on flexible maximum power limit, each BS has the freedom to allocate more power to a specific RB, under the constraint that on the average the total power allocation must not exceed the limit set by the RNC allocation. With this strategy, the allocation at the BS level is more flexible, since the BS has the ability to increase further the amount of power in order to meet the SINR threshold on a specific RB. Nevertheless, the BSs have to keep track of the excessive power allocations and try to absorb them in consequent RBs power allocation.

According to at least some embodiments, moving average should be used as it would make it easier for the BS to fulfill the desired constraint imposed by the RNC towards the end of the super frame. Other averaging techniques or metrics (e.g. percentile) could also be used by the base station that could satisfy the RNC restriction in terms of power allocation on super-frame level.

The flexible power constraint strategy may be more aggressive and may allow the base station to better exploit opportunistic scheduling schemes.

Overview of a Communication System

In FIG. 6 an overview of a communication system 600 for resource allocation according to some embodiments is presented. Said system may comprises a base station 602 for distributed transmission power allocation, a radio network controller 604 for distributed transmission power determination, and at least one use equipment 606.

The radio network controller 604 may comprise a regulator function 610, which is connected to a power determination unit 612. The base station 602 may comprise a gain unit control 606 and a power allocation unit 608, whereas the user equipment 606 may comprise a gain unit 614 as well as a radio channel unit 616.

These units and their functions may be comprised in other units having a multiple functions. FIG. 6 presents a system overview according to some embodiments only.

In FIG. 6, the average gain matrix G may be provided from the gain unit control 606 of the base station 602 to the power determination unit 612. From the gain unit control 606 of the base station 602, the matrix Q showing the length of each user's buffer, that is the queue length or buffer state, may be provided to the regulator function 610 of the radio network controller 604. T_previousbeing the achieved throughput per user for the foregoing super-frame, in other words how many bits each user has received during the last super-frame, may also be provided by the base station 602 to the regulator 610. The type of service S per user, being either best effort, guaranteed bit rate, applicable for Narrowband, or elastic, may similarly be provided by the base station 602, to the regulator function 610 of the radio network controller 604.

T_reqbeing the required throughput per user in order to satisfy its quality of service (QoS) constraints, which can be in the form of actual bit rate or number of RBs per user, may be provided by the regulator function 610 to the power determination unit 612. The transmission power level matrix P for long-term power allocation may be provided by the power determination unit 612 to the power allocation unit 608 of the base station 602.

The regulator 610 may be an embedded function in the radio network controller entity which aims on regulating/deciding the number of resource blocks (RBs) per user (or the bit rate for the upcoming super-frame period) in order to keep users of the UEs satisfied in terms of their QoS requirements.

The resource block decisions may be based on the queue length, the achievable throughput during the foregoing super-frame period and the service type. Based on these measurements the regulator 610 may estimate the required throughput per user (or number of RBs) for the upcoming super-frame.

In addition, instantaneous gain data or an average gain report may be provided by the gain unit 614 of the user equipment 606.

According to at least some embodiments of the present invention the user equipment may comprise a mobile phone. The user equipment may alternatively also comprise a personal computer or a personal digital assistant device.

The radio channel unit 618 may receive the downlink data traffic comprising pilot signals from the power allocation unit 608 of the base station 602. Also, instantaneous report such as channel quality indicators may be provided by the radio channel unit 618 of the UE to the power allocation unit 608 of the base station 602, while taking a quick look at FIG. 7 presenting a signaling scheme according to some embodiments of the present invention, which is presented in more detail down below.

Message Sequence Diagram According to Some Embodiments

Referring to FIG. 7, showing a message sequence diagram according to at least some embodiments.

In addition to messages exchanged between UEs and BSs, new messages are established between the BSs and the RNC, according to some embodiments of the present invention. At the beginning of a super-frame, BSs transmit pilot signals, step 702, which are then used by each UE to estimate the average over fast fading channel gains, between itself and the rest BSs of the cluster controlled by the RNC. The average over fast fading channel gains is commonly termed long term average as it filters out the effect of fast fading or small scale fading.

The average channel gain estimations may then be reported back, step 704, to the connected BS. Next, each BS may report these gain values, step 706, which basically correspond to the rows of the aforementioned G matrix, as well as the performance requirements of each user to the RNC. Now, RNC is able to construct the gain matrix G and can then calculate a close to optimum power level, step 708, which when allocated would achieve a close to optimal long-term data bit throughput of the system. The resulted P matrix is then distributed to all BSs, step 710. Each BS transmits pilot signals over the allowed set of RBs, step 712, as it was defined by the RNC. These signals are used for the estimation of the instantaneous channel gains by the UEs, which are then reported to their serving base stations, step 714. Based on these measurements as well as on the power allocation matrix P, each BS performs scheduling and power allocation, step 716, over the allowed set of RBs according to the strategies presented in the previous section. Thereafter, data along with new pilot signals can the transmitted to the UEs using the allocated power, step 718.

The received pilot signals are then used for the estimation of the instantaneous channel gains by the UEs, which again are reported as channel quality indicators (CQI) for instance, to their serving base stations in step 720. Again, based on these measurements as well as on the power allocation matrix P, each BS performs a novel scheduling and power allocation, step 722, over the allowed set of RBs according to the strategies as mentioned above. Now, data along with updated pilot signals can the transmitted to the UEs using the allocated power, step 724.

This way of sending downlink traffic along with pilot signals and receiving instantaneous reports are continued on the frame level, step 726. When the current super-frame is coming to its end, an average gain report may be transmitted, step 728, from the UE to the service BS, as input data for an updated transmission power level determination on the super-frame level, whereafter the steps as described above may be scheduled, such as steps 706 and onwards.

Applicability to Uplink Channel and Power Allocation

The description in the preceding sections considers only downlink scenario. However, the invention is also applicable for mitigating inter-cell interference in the uplink. Similarly to downlink, for uplink scenario the RNC will carry out the long term uplink power and channel allocation (uplink resource block) by using the same parameters and principles described in the preceding sections. The short term uplink resource allocation is also done at the base station, which would therefore use methodologies similar to those described above. The current LTE system design allows the possibility to perform channel dependent scheduling also in the uplink. This is due to the fact that UE could be configured to send uplink reference signals, so called sounding reference signals, at least sparsely over the entire band. Thus every time, during a super-frame, the UE is scheduled by the base station is also informed about maximum allowed uplink power limit per RB. This will ensure that the long term conditions set by the RNC are fulfilled and will eventually minimize the uplink inter-cell interference.

It should be understood that the presented embodiments of the present invention are only a few examples of the variety of embodiments that are comprised within the present invention.

It is emphasized that the present invention can be varied in many ways, of which the alternative embodiments as presented are just a few examples. These different embodiments are hence non-limiting examples.

The radio network controller may be comprised within one entity, but may alternatively also be comprised of distributed functions with one or several base stations.

Advantages of at Least Some of the Embodiments

Mitigates interference in the neighbouring cells and allows the system to operate with frequency reuse-1, which achieves a more efficient usage of the bandwidth.

Quality targets on the short term, i.e. on the frame level and on the long term, i.e. on the super frame level, can be achieved.

The user bit rate performance requirements are achieved.

The power and channel resources can be used more efficiently.

The optimization scheme is simple in the sense that it makes use of existing measurements.

The scheme is applicable in both uplink and downlink transmission.

Method and System for Mitigating Inter-Cell Interference

Information

Publication Number

Date Filed

Date Published

Inventors

CPC

US Classifications

International Classifications

Abstract

Description

Claims

PCT Information

Provisional Applications (1)