Wireless telecommunications network

Abstract

A wireless telecommunications network, such as a multi-user Multiple-Input-Multiple-Output (MIMO) system, has a transmitter unit, such as base station (110 fig.1), having a plurality of transmitters a

Description

Field of the Invention The present invention relates to a wireless telecommunications network. Background A wireless telecommunications network may use a Multiple-Input-Multiple-Output (MIMO) technology to improve capacity. The capacity may be defined as a “sum capacity” (or “sum rate”) being the maximum aggregate data rate of all users. It is desirable to maximise the sum capacity of the multi-user MIMO wireless telecommunications network. A multi-user MIMO wireless telecommunications network may have a limitation of the number of users that may be communicated with simultaneously. For example, when the multi-user MIMO wireless telecommunications network uses Zero-Forcing BeamForming (ZFBF) in which a transmitting node has M transmitters, the number of users that may be communicated with simultaneously should not exceed the number of transmitters. In order to maximise the sum capacity of the multi-user MIMO wireless telecommunications network in which there is a limitation of the number of users that may be communicated with simultaneously, a particular subset of users (known as the “active set of users”) is selected for communication. A process, known as user scheduling, selects the active set of users that maximises the sum capacity of the multiuser MIMO wireless telecommunications network based on, for example, inter-user interference and channel conditions between the transmitting node and each user. Summary of the Invention According to a first aspect of the invention, there is provided apparatus for controlling a transmitter unit in a wireless telecommunications network, wherein the transmitter unit has a plurality of transmitters and the wireless telecommunications network has a plurality of receiver units, the apparatus comprising: a processor configured to: define a first set of the plurality of receiver units comprising the plurality of receiver units, iteratively add a receiver unit of the first set of the plurality of receiver units to a second set of the plurality of receiver units until a termination condition is met by: determining a channel metric for each receiver unit of the first set of the plurality of receiver units, wherein the channel metric for each receiver unit is based on a channel vector of the receiver unit of the first set of the plurality of receiver units and a channel error vector of the receiver unit, adding a receiver unit of the first set of the plurality of receiver units having the greatest determined channel metric of the first set of the plurality of receiver units to the second set of the plurality of receiver units, and redefining the first set of the plurality of receiver units, in which a correlation between a respective channel vector of each receiver unit in the redefined first set of the plurality of receiver units and the channel vector of the receiver unit added to the second set of the plurality of receiver units in a current iteration satisfies an orthogonality threshold, and cause the transmitter unit to transmit to the determined second set of the plurality of receiver units. According to a second aspect of the invention, there is provided a method of controlling a transmitter unit in a wireless telecommunications network, wherein the transmitter unit has a plurality of transmitters and the wireless telecommunications network has a plurality of receiver units, the method comprising the steps of: defining a first set of the plurality of receiver units comprising the plurality of receiver units, iteratively adding a receiver unit of the first set of the plurality of receiver units to a second set of the plurality of receiver units until a termination condition is met by: determining a channel metric for each receiver unit of the first set of the plurality of receiver units, wherein the channel metric for each receiver unit is based on a channel vector of the receiver unit of the first set of the plurality of receiver units and a channel error vector of the receiver unit, adding a receiver unit of the first set of the plurality of receiver units having the greatest determined channel metric of the first set of the plurality of receiver units to the second set of the plurality of receiver units, and redefining the first set of the plurality of receiver units, in which a correlation between a respective channel vector of each receiver unit in the redefined first set of the plurality of receiver units and the channel vector of the receiver unit added to the second set of the plurality of receiver units in a current iteration satisfies an orthogonality threshold; and cause the transmitter unit to transmit to the determined second set of the plurality of receiver units. In the first or second aspects, the channel vector for each receiver unit of the first set of the plurality of receiver units may be based on a channel vector of the receiver unit of the first set of the plurality of receiver units in a timeslot occurring D timeslots prior to a current timeslot. In the first or second aspects, the channel error vector of the receiver unit of the first set of the plurality of receiver units may represent an error between a channel vector of the receiver unit of the first set of the plurality of receiver units in the current timeslot and the channel vector of the receiver unit of the first set of the plurality of receiver units in the timeslot occurring D timeslots prior to a current timeslot. In the first or second aspects, the channel metric for each receiver unit of the first set of the plurality of receiver units may be further based on a channel coefficient. In the first or second aspects, the channel metric of each receiver unit of the first set of the plurality of receiver units may be determined as: 1 + (X - l)p||e[n]||2 in which: • p is defined as p in which Pt is the maximum transmission power of the transmitter unit; • 5, is defined as pp2, in which p is a correlation coefficient; • h[n - D] is the channel vector of the receiver unit of the first set of the plurality of receiver units in a timeslot occurring D timeslots prior to a current timeslot; • Ks is a configurable threshold representing a maximum size of the second set of receiver units; and • e[n] is the error vector of the receiver unit of the first set of the plurality of receiver units in the current timeslot. In the first or second aspects, the correlation between the respective channel vector of each receiver unit in the redefined first set of the plurality of receiver units and the channel vector of the receiver unit added to the second set of the plurality of receiver units in the current iteration may relate to the respective channel vector in the timeslot occurring D timeslots prior to the current timeslot of each receiver unit in the redefined first set of the plurality of receiver units and may further relate to the channel vector in the timeslot occurring D timeslots prior to the current timeslot of the receiver unit added to the second set of the plurality of receiver units in the current iteration. In the first or second aspects, the termination condition may be one or more of: a size of the second set of the plurality of receiver units equalling a configurable maximum size of the second set of receiver units, and the redefined first set of the plurality of receiver units being a null set. In the first or second aspects, the maximum size of the second set of receiver units may be defined as a ratio of a count of the plurality of receiver units to a constant, the constant may have a value equal to or greater than 4. In the first or second aspects, the orthogonality threshold may be between 0.15 and 0.8, or more specifically between 0.25 and 0.8. According to a third aspect of the invention, there is provided a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method of the first aspect of the invention. The computer program may be stored on a computer-readable data carrier. Brief Description of the Figures In order that the present invention may be better understood, embodiments thereof will now be described, by way of example only, with reference to the accompanying drawings in which: Figure 1 is a schematic diagram of a wireless telecommunications network; Figure 2 is a flow diagram illustrating a first method; Figure 3 is a flow diagram illustrating a second method; and Figure 4 is a graph illustrating sum rate against orthogonality threshold for a conventional user scheduling process and the methods of Figures 2 and 3. Detailed Description Figure 1 illustrates a wireless telecommunications network 100 having a base station 110. The base station 110 has a plurality of transceivers that are configured to communicate in a single coverage area (that is, a single “cell”). The wireless telecommunications network 100 also includes a plurality of User Equipment, UE, 120, positioned within the coverage area of the base station 110. Each UE of the plurality of UE 120 has a single transceiver for communication with the base station 110. The wireless telecommunications network 100 is therefore an example of a multi-user Multiple-Input-Single-Output (MISO) system. The following description describes a communication model for the wireless telecommunications network 100 to aid understanding of a method of user scheduling in the wireless telecommunications network 100 (as described later in the description). Considering a downlink communication from the base station 110 (having M transceivers) to Ks UE, the transmitted downlink signal, x, may be represented as: k=l In which: • Ks is the active set of UE (in which Ks « M); • sk represents the symbol for UE k (k = 1, ..., Ksy • wk is an M x 1 precoding vector; and • pk is the allocated power to UE k. The received signal at UE k, yk, is: _ w s + V ^hws +n yk -\]Pk"-k^k^k ' / <' ^k j = l,j*k (1) In which: • hk is a channel vector for the channel between the base station 110 and user k, which may be represented as a 1 x M vector with independently and identically distributed (i.i.d) zero-mean, unit variance complex Gaussian elements; and • nk e C(0,l) denotes the complex additive white Gaussian noise (AWGN) random variable. The achievable downlink sum rate, SumRate, of the wireless telecommunications network 100 with precoding is: SumRate = ks £log2(l k=l Ks + SINRk) =^log2 k=l Pk\hkwk\2 1+^=i,^k / pj\hkwi\2 (2) In which: • EfciillPfcWfcH2 <Pt, and Pt is the maximum transmit power of the base station 110. A channel estimation process is implemented to enable each UE of the plurality of UE 120 to estimate the channel between the base station 110 and UE. Each UE reports the channel estimate, such as Channel State Information (CSI), to the base station 110. Once the base station 110 has received the CSI of all users, it may implement Zero Force BeamForming (ZFBF) to remove interference between UE of the active set of UE (such that hkWj = 1, for j * k). To find the orthogonal channels between the base station 110 and the UE, aZF pre-processing matrix, W, obtains the pseudo-inverse of the aggregate channel of the active set of UE, H, according to: w = hh(hhhy1 In which the beamforming vectors are normalised so as to have unitary norm. These aspects of the communication model are discussed in “Multi-Antenna Downlink Channels with Limited Feedback and User Selection”, Yoo et al., IEEE Journal on Selected Areas in Communications, Vol. 25, No. 7, September 2007 (hereby incorporated by reference). A stationary ergodic Gauss-Markov process is used to model the time-variations of the channel between the base station 110 and UE of the plurality of UE 120. Based on this model, the channel elements are constant for one symbol period while the change from one symbol to the Dth subsequent symbol is modelled as: hk [n] = phk [n - D] + ek [n] (3) In which: • hk[n] is the channel vector at time slot n; • p is the correlation coefficient using the classical Clarke’s isotropic scattering model; • hk[n - D] is the channel vector at time slot n- D, uncorrelated with efc[n] • D is the number of symbol periods, being greater than 1, between iterations of the first and second method described below; and • ek[n] denotes an error vector with i.i.d. entries given by [n]~CW(0, E2 / ), in which £2 is the variance of the error vector (detailed below) and I is the identity matrix. The error vector, ek[n], represents a difference between the channel vector in the timeslot [n] and the correlated version of channel vector in the timeslot [n-D], The variance of the error vector is given by: E2 = 1 — p2 (4) The correlation coefficient, p, is a statistical measure of the degree to which changes to the value of hk[n] predict change to the value of hk[n - D] and may be obtained as p = J0(2nfdDTs\ in which: • Jo(') represents the zeroth-order Besel function of the first kind; • fd is the Doppler spread; and • Ts is the symbol period. These aspects of the communication model are discussed in “Performance Analysis of Closed-Loop Transmit Diversity in the Presence of Feedback Delay”, Onggosanusi et al., IEEE Transactions on Communications, Vol. 49, No. 9, September 2001 (hereby incorporated by reference). A first method will now be described with reference to Figure 2. The first method is implemented by each UE of the plurality of UE 120 (and will be described in the context of a first UE of the plurality of UE 120). In a first step, S101, the first UE performs a channel estimation process to determine the channel vector of the channel from the base station 110 to the first UE in the current time slot, n. This is stored in memory as [n] and associated with a timestamp representing the current timeslot, n, at which the channel vector was determined. The first UE also stores, in memory, at least one additional channel vector (i.e. h^n- D]) being the channel vector of the channel from the base station 110 to the first UE occurring D timeslots prior to the current timeslot n. The first UE also stores a value for the correlation coefficient, p, in memory. In step S103, the first UE calculates the error vector, ejn], based on equation 3 above and the stored values for ^[n], h^n - D] and p. In step S105, the first UE sends a message to the base station 110 to report the values for ^[n] and e1 [n]. The process is repeated every D-th timeslot. A second method will now be described with reference to Figure 3. This second method is a method of user scheduling to determine an active set of UE for downlink communications. As noted above, each UE of the plurality of UE 120 estimate and report their respective channel vector and respective error vector to the base station 110. The channel estimation and reporting processes are implemented periodically by each UE of the plurality of UE 120 (i.e. every D timeslots) so as to reduce the feedback overhead in the uplink channels from the plurality of UE 120 to the base station 110 (relative to providing this feedback in every timeslot). That is, by increasing the delay between these periodic updates (such that D is greater than 1), there is a reduction in the feedback overhead. However, this “delayed CSI at the Transmitter” (“delayed CSIT”), when combined with ZFBF, may result in inter-user-interference which reduces the Signal to Interference plus Noise Ratio (SINR) of one or more users and therefore reduces the sum-rate of the wireless telecommunications network 100. The following method provides a trade-off between increasing the number of users in the system (thus increasing multiuser diversity) and decreasing CSI feedback overhead. The second method includes an initialisation step, S201, in which the base station 110 defines an active set of UE, S, a candidate set of UE, V, and a maximum size, Ks, of the active set of UE. The active set of UE, S, is initialised as a null (i.e. empty) set and the candidate set of UE, V, is initialised as a set containing all UE of the plurality of UE (i.e. V = [1,... , / <]). The maximum size, Ks, of the active set of UE So is initialised as in which M equals the number of UE in the plurality of UE 120 and a is equal to 4. The following steps iteratively add UE from the candidate set of UE, V, to the active set of UE, S, until a termination condition is met. In step S203, the base station 110 receives the respective channel vector, hk[n], and respective error vector efe[n] from each UE of the plurality of UE 120 (as described above in the first method). These are stored in memory. The channel vectors are each associated with the timestamp representing the time slot in which the channel vector was determined. The base station 110 stores a series of channel vector values for each UE of the plurality of UE 120 (including at least the channel vector hk[n - D] of a timeslot occurring D timeslots prior to the current timeslot, ri). In step S205, the base station 110 determines an SINR lower bound (“SLB”) value for each UE of the candidate set of UE, V. The SLB value for each UE is determined as a function of the channel vector, hv[n - D], for the UE in the timeslot occurring D timeslots prior to the current timeslot, n, and the error vector, ev[n], for the UE in the current timeslot, n. Specifically, the SLB value for UE k is calculated as: eIR = (5) v 1 + (Ks - l)p\\ev[nW In which is equal to pp2, in which p is a correlation coefficient, and p = —. Ks In step S207, the base station 110 identifies a UE (hereinafter denoted UE n) of the candidate set of UE, V, having the greatest SLB value and adds UE n to the active set of UE, S. In step S209, the base station 110 updates the candidate set of UE so as to only include UE of the plurality of UE 120 that have a respective channel vector hk[n-D] in the timeslot occurring D timeslots prior to the current timeslot, n, that satisfies an orthogonality threshold, eh, with the channel vector of the identified UE, h„[n - D] in the timeslot occurring D timeslots prior to the current timeslot, n. That is, k e Vi-lrk £ ?r(0, (6) In which: • = h(W(0)[n-D], and • eh is a configurable parameter having a value between 0 and 1, preferably between 0.1 and 0.8, and more preferably between 0.25 and 0.8. In step S211, the base station 110 determines whether one or more termination criteria has / have been met. In this example, the termination criteria include: • the updated candidate set of UE is a null set (i.e. no UE of the plurality of UE 120 have a respective channel vector hk[n - D] in the timeslot occurring D timeslots prior to the current timeslot, n, that satisfies an orthogonality threshold, eh, with the channel vector of the identified UE, hn[n-D] in the timeslot occurring D timeslots prior to the current timeslot, n); and • the number of UE in the active set of UE, S, equals the maximum size, Ks, of the active set of UE, S. If no termination criteria are met, then the method loops back for a further iteration of steps S205 to S211, in which each iteration is based on the candidate set of UE as updated in step S211 of the previous iteration. Once terminated, the active set of UE, S, is complete and the base station 110 is configured (in step S213) to serve the complete active set of UE, S. Following expiry of D timeslots, the second method is repeated by looping back to step S201. The subsequent iteration of the second method is therefore repeated following expiry of the next D timeslots at which point the base station 110 receives new values for hk[n] and ek[n] following performance of the first method by each UE of the plurality of UE 120. The active set of UE, S, is therefore defined for D timeslots. Figure 4 is a graph illustrating the average sum rate against the orthogonality threshold of a first and second simulation of the user scheduling process described above using ZFBF and equal power allocation. Figure 4 also illustrates, for the purposes of comparison, the average sum rate against the orthogonality threshold of a third and fourth simulation of a Near-Orthogonal User Scheduling (“NEOUS”) process described in “Sum rate optimal multi-antenna downlink beamforming strategy based on clique search,”, T. Yoo et al., IEEE Globecom, Dec 2005 (hereby incorporated by reference) and “Asymptotic analysis of SDMA systems with near-orthogonal user scheduling (NEOUS) under imperfect CSIT,” V. K. N. Lau, IEEE Trans. Commun., vol. 57, no. 3, pp. 747-753, Mar. 2009 (hereby incorporated by reference), again using ZFBF and equal power allocation. The first and third simulations use a maximum size, Ks, value of 30 and the second and fourth simulations use a maximum size, Ks, value of 20. Figure 4 illustrates that the user scheduling process of Figure 3 outperforms the NEOUS process by achieving improved sum rate values once the orthogonality threshold is greater than a certain value (around 0.15 in Figure 4). This is due to the user scheduling process of Figure 3 determining an active set of UE based on an SLB value of each UE (being a function of the channel vector between the base station 110 and the UE in the timeslot occurring D timeslots prior to the current timeslot, n, and the error vector for the UE in the current timeslot, n), instead of the channel power gain as used in the NEOUS process, whilst still ensuring each UE of the active set of UE satisfies an orthogonality threshold. The following is noted regarding the SLB. The channel vector between the base station 110 and UE k of the plurality of UE 120 may be considered a random variable. The SI NR for UE k, SINRk, may then also be defined as a random variable. The expected value for the SINR for UE k, SINRk, may therefore be calculated as: SINRk + SINRk +--- + SINR. E{SINRk] =---- Where n = 1,..., N. A lower bound of a set is defined as a value that is less than or equal to every element of the set. Using Jensen’s inequality, we have SINRk >E{SINRk}. Therefore, the SLB may therefore be derived as: (1 + P SJik| ek [n] Wj [n] |2 ||2 - 1 + (Ks - l)p||ev[n]||2 The following is noted regarding the orthogonality threshold. If the value of the orthogonality threshold is too high (e.g. more than 0.8), then the second method described above defines an active set of UE within which there is a large correlation between UE. This can reduce the sum rate due to inter-user-interference. If the value of the orthogonality threshold is too low (e.g. less than 0.25), then the active set of UE may be small (that is, the chances of the candidate set of UE being a null set is increased), resulting in a low sum rate. The orthogonality threshold may therefore be configured as a value between these two extremes, which may be determined by simulation or calibration. The processes described above relate to a MISO wireless telecommunications network in which the base station 110 communicates in a single cell. However, the skilled person will understand that these characteristics are non-essential, and the processes may be applied to other forms of wireless telecommunications networks, such as a Multiple-Input Multiple-Output (MIMO) system, and may also apply to multi-cell base stations such as a tri-sector base station or a base station implementing beamforming. The processes described above also relate to a base station defining an active user set for downlink communications. However, this is also non-essential. The processes may be implemented by any transmitting unit (having a plurality of transmitters) in any wireless telecommunications network that concurrently (i.e. simultaneously) communicates with a plurality of receiving units. These communications may be downlink or uplink, or may be between equivalent network nodes. The skilled person will also understand that it is non-essential that the second method is performed by the transmitting unit itself, and instead may be performed by another entity that controls the transmitting unit. In the second method described above, the maximum size of the active user set, Ks, is defined as p in which M equals the number of UE in the plurality of UE 120 and a is equal to 4. The value of a is a configurable parameter and is preferentially set to 4 as linear precoding is almost optimal at this value. However, this is non-essential. That is, the sum rate of the system improves with increasing a, but increasing a increases the cost and complexity of the transmitting unit and decreases the number of UE being transmitted to. The value of a may therefore be equal to or greater than 4. The skilled person will understand that any combination of features is possible within the scope of the invention, as claimed.

Claims

1. Apparatus for controlling a transmitter unit in a wireless telecommunications network, wherein the transmitter unit has a plurality of transmitters and the wireless telecommunications network has a plurality of receiver units, the apparatus comprising: a processor configured to:define a first set of the plurality of receiver units comprising the plurality of receiver units,iteratively add a receiver unit of the first set of the plurality of receiver units to a second set of the plurality of receiver units until a termination condition is met by:determining a channel metric for each receiver unit of the first set of the plurality of receiver units, wherein the channel metric for each receiver unit is based on a channel vector of the receiver unit of the first set of the plurality of receiver units and a channel error vector of the receiver unit,adding a receiver unit of the first set of the plurality of receiver units having the greatest determined channel metric of the first set of the plurality of receiver units to the second set of the plurality of receiver units, andredefining the first set of the plurality of receiver units, in which a correlation between a respective channel vector of each receiver unit in the redefined first set of the plurality of receiver units and the channel vector of the receiver unit added to the second set of the plurality of receiver units in a current iteration satisfies an orthogonality threshold, andcause the transmitter unit to transmit to the determined second set of the plurality of receiver units.

2. Apparatus as claimed in Claim 1, wherein the channel vector for each receiver unit of the first set of the plurality of receiver units is based on a channel vector of the receiver unit of the first set of the plurality of receiver units in a timeslot occurring D timeslots prior to a current timeslot.

3. Apparatus as claimed in Claim 1 or Claim 2, wherein the channel error vector of the receiver unit of the first set of the plurality of receiver units represents an error between a channel vector of the receiver unit of the first set of the plurality of receiver units in the current timeslot and the channel vector of the receiver unit of the first set of the plurality of receiver units in the timeslot occurring D timeslots prior to a current timeslot.

4. Apparatus as claimed in any one of the preceding claims, wherein the channel metric for each receiver unit of the first set of the plurality of receiver units is further based on a channel coefficient.

5. Apparatus method as claimed in any one of the preceding claims, wherein the channel metric of each receiver unit of the first set of the plurality of receiver units is determined as:OMn-DHI2 1 + (Ks - l)p||e[n]||2 in which:• p is defined as — in which Pt is the maximum transmission power of the Kstransmitter unit;• is defined as pp2, in which p is a correlation coefficient;• h[n - D] is the channel vector of the receiver unit of the first set of the plurality of receiver units in a timeslot occurring D timeslots prior to a current timeslot;• is a configurable threshold representing a maximum size of the second set of receiver units; and• e[n] is the error vector of the receiver unit of the first set of the plurality of receiver units in the current timeslot.

6. Apparatus as claimed in any one of the preceding claims, wherein the correlation between the respective channel vector of each receiver unit in the redefined first set of the plurality of receiver units and the channel vector of the receiver unit added to the second set of the plurality of receiver units in the current iteration relates to the respective channel vector in the timeslot occurring D timeslots prior to the current timeslot of each receiver unit in the redefined first set of the plurality of receiver units and further relates to the channel vector in the timeslot occurring D timeslots prior tothe current timeslot of the receiver unit added to the second set of the plurality of receiver units in the current iteration.

7. Apparatus as claimed in any one of the preceding claims, wherein the termination condition is one or more of:a size of the second set of the plurality of receiver units equalling a configurable maximum size of the second set of receiver units, andthe redefined first set of the plurality of receiver units being a null set.

8. Apparatus as claimed in Claim 7, wherein the maximum size of the second set of receiver units is defined as a ratio of a count of the plurality of receiver units to a constant, the constant having a value equal to or greater than 4.

9. Apparatus as claimed in any one of the preceding claims, wherein the orthogonality threshold is between 0.15 and 0.8.

10. Apparatus as claimed in Claim 9, wherein the orthogonality threshold is between 0.25 and 0.

811. A method of controlling a transmitter unit in a wireless telecommunications network, wherein the transmitter unit has a plurality of transmitters and the wireless telecommunications network has a plurality of receiver units, the method comprising the steps of:defining a first set of the plurality of receiver units comprising the plurality of receiver units,iteratively adding a receiver unit of the first set of the plurality of receiver units to a second set of the plurality of receiver units until a termination condition is met by:determining a channel metric for each receiver unit of the first set of the plurality of receiver units, wherein the channel metric for each receiver unit is based on a channel vector of the receiver unit of the first set of the plurality of receiver units and a channel error vector of the receiver unit,adding a receiver unit of the first set of the plurality of receiver units having the greatest determined channel metric of the first set of the plurality of receiver units to the second set of the plurality of receiver units, andredefining the first set of the plurality of receiver units, in which a correlation between a respective channel vector of each receiver unit in the redefined first set of the plurality of receiver units and the channel vector of the receiver unit added to the second set of the plurality of receiver units in a current iteration satisfies an orthogonality threshold; and cause the transmitter unit to transmit to the determined second set of the plurality of receiver units.

12. A method as claimed in Claim 11, wherein the channel vector for each receiver unit of the first set of the plurality of receiver units is based on a channel vector of the receiver unit of the first set of the plurality of receiver units in a timeslot occurring D timeslots prior to a current timeslot.

13. A method as claimed in Claim 11 or Claim 12, wherein the channel error vector of the receiver unit of the first set of the plurality of receiver units represents an error between a channel vector of the receiver unit of the first set of the plurality of receiver units in the current timeslot and the channel vector of the receiver unit of the first set of the plurality of receiver units in the timeslot occurring D timeslots prior to a current timeslot.

14. A method as claimed in any one of Claims 11 to 13, wherein the channel metric for each receiver unit of the first set of the plurality of receiver units is further based on a channel coefficient.

15. A method as claimed in any one of Claims 11 to 14, wherein the channel metric of each receiver unit of the first set of the plurality of receiver units is determined as:<Mh[n-£>]||2 1 + (Ks - l)p||e[n]||2 in which:• p is defined as — in which Pt is the maximum transmission power of the Kstransmitter unit;• is defined as pp2, in which p is a correlation coefficient;• h[n - D] is the channel vector of the receiver unit of the first set of the plurality of receiver units in a timeslot occurring D timeslots prior to a current timeslot;• Ks is a configurable threshold representing a maximum size of the second set of receiver units; and• e[n] is the error vector of the receiver unit of the first set of the plurality of receiver units in the current timeslot16. A method as claimed in any one of Claims 11 to 15, wherein the correlation between the respective channel vector of each receiver unit in the redefined first set of the plurality of receiver units and the channel vector of the receiver unit added to the second set of the plurality of receiver units in the current iteration relates to the respective channel vector in the timeslot occurring D timeslots prior to the current timeslot of each receiver unit in the redefined first set of the plurality of receiver units and further relates to the channel vector in the timeslot occurring D timeslots prior to the current timeslot of the receiver unit added to the second set of the plurality of receiver units in the current iteration.

17. A method as claimed in any one of Claims 11 to 16, wherein the termination condition is one or more of:a size of the second set of the plurality of receiver units equalling a configurable maximum size of the second set of receiver units, andthe redefined first set of the plurality of receiver units being a null set.

18. A method as claimed in Claim 17, wherein the maximum size of the second set of receiver units is defined as a ratio of a count of the plurality of receiver units to a constant, the constant having a value between 3 and 5.

19. A method as claimed in any one of Claims 11 to 18, wherein the orthogonality threshold is between 0.15 and 0.8.

20. A method as claimed in Claim 19, wherein the orthogonality threshold is between 0.25 and 0.8.

21. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of any one of Claims 11 to 20.

22. A computer-readable data carrier having stored thereon the computer program of claim 21.

Images