Page 101 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 6 – Wireless communication systems in beyond 5G era

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 6

Fig. 8 – Results of 10-cluster partitioning based on
Fig. 7 – Experimental environment experimental data
was exceeded by nearly all of the combinations. A conceptual diagram of the experiment is shown in
However, as far as the best value is concerned, the Fig. 7. The experiment was conducted in a
proposed method asymptotically approached the k- laboratory within the NTT Yokosuka R&D Center.
means method. The reason we give for this is that The floor of the laboratory was laid with 450-mm-
the distribution of amplitudes among the clusters square tiles and an 18 × 18 arrangement of these
could be kept fixed even for clustering by the k- tiles were used in the experiment to secure an
means method due to an increase in the number of experimental space of 324 tiles in total. To indicate
cluster partitions. the positions of the experimental equipment, the
4.2 Number of clusters and performance upper left corner of the diagram in Fig. 7 was taken
to be the origin with the x-axis and y-axis extending
Next, to clarify the relationship between the in rightward and downward directions,
number of cluster partitions and channel capacity in respectively. To apply the model shown in Fig. 2 to
the proposed method and to compare results this space, we fixed the sender (Tx) at position (1, 9),
obtained by the proposed method with those of an the interfered receiver (Ix) at position (18, 6), and
exhaustive search with no partitioning into clusters, the receiver (Rx) at position (18, 12) while enabling
we evaluated the channel capacity of receiver D an IRS reflector (R) to be set up at all positions from

when partitioning a total of 20 IRS reflectors into 5, (1, 1) to (18, 18). In the experiment, a 5 GHz signal
10, 15, and 20 (exhaustive search) clusters. Results generator (20 MHz bandwidth) specified by IEEE
are shown in Fig. 6. It can be seen that the 802.11ac was set up on a tile and treated as an IRS
performance of the proposed method increasingly reflector and the generated signal was measured by
deviated from that of the exhaustive search as the an oscilloscope (Keysight Technologies) set up at
number of clusters decreased but that dividing the the positions of the sender, interfered receiver, and
20 IRS reflectors into half that number of clusters, receiver. This operation was repeated to obtain the
i.e., 10 clusters, could secure a channel capacity of channel state information for all 324 locations
about 70% that of the exhaustive search. where the IRS reflector was set up.
Now, on dividing these 324 channels obtained by
5. EXPERIMENTAL EVALUATION
experiment into 10 clusters and entering the
The characteristics of the proposed method depend representative value of each into the solver, we
greatly on the propagation channel, so evaluation evaluated the CDF of channel capacity obtained

based on actual channel measurements is essential by receiver D (see Eq. (12)) the same as in
for practical implementation. To therefore clarify its subsection 4.2. Results are shown in Fig. 8. Similar
effectiveness in an actual environment, we to the results of Fig. 5 obtained by simulation, these
measured an actual propagation channel using an experimental results also showed that the proposed
oscilloscope and performed a simulation treating method significantly increased the worst value for
the measured values as a propagation channel channel capacity from 0 bit/s/Hz to 2.4 bit/s/Hz
passing through IRS reflectors. compared with two conventional methods thereby
demonstrating its effectiveness in improving
propagation characteristics.

© International Telecommunication Union, 2021 89