Page 60 - 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

3.3 Beam selection algorithm based on
ℎ ( UE , , )
BS
UR R ⋯R 1 B routing
−1

= ℎ ( , ) ℎ ( ) (16)
UR UE R R
−1
∙ ⋯ ℎ ( , ) In order to maximize channel capacity, the

R 1 B BS
optimized routes and corresponding beams of RSs
where , , ⋯ = 1, … , is the RS index in K-1th need to be derived through full-search, which
RS
hop and K hop. Although ≠ ≠ ⋯ ≠ to avoid the traverses all the combinations of the beam
self-loops and to avoid multiple use of the same RS selections in the BS and RSs. Assuming that signals
in different paths, those constraints are applied by are relayed in a maximum number of K hops and the
the routing algorithm in Section 3.3. number of available RSs is , the computational
RS
By summing up the above derivations, the channel cost of such full-search algorithm max is described
response between the BS and UE, with as follows.
considerations of all paths including self-loops,
denoted by ℎ URB ( UE , , ) , can be defined as = ∑( 2 ) (19)

BS
follows. max RS RS BS
=1
ℎ URB ( , , ) = ℎ UB ( UE , , )
UE
BS
BS
where is the number of candidate beams of RS
RS RS
for both sides .
+ ∑ ℎ ( , , )
UR 1 B UE BS
=1 It is obviously indicated by Eq. (19) that the full-
search algorithm inevitably results in an enormous
(17)
RS RS and unaffordable computational cost. Since the
+ ∑ ∑ ℎ ( , , ) + ⋯ position of the UE could change time by time, the
UR R 1 B UE BS
2
=1 =1 routes and beams need to be recalculated
instantaneously according to the new UE position to
+ ∑ ⋯ ∑ ℎ ( UE , , ) guarantee the mmWave communication. Therefore,
BS
⏟ UR R ⋯R 1 B
−1
it is neither practical nor feasible to use the full-
search method for fast moving UEs such as vehicles.
where ℎ UB ( UE , , ) is the channel response of
BS
the direct path between the BS and UE. In order to complete the selection of beams in real
time, we propose an algorithm to determine relay
3.2 Noise vector with multi-hop analog relay routes and corresponding beams of RSs
The UE × 1 noise vector from 1st to K-th hop RSs sequentially by finding the shortest paths.
( ) is defined as follows. Fig. 6 shows the proposed routing algorithm for the
RS
massive multi-hop relay MIMO system. It is
( )
RS
assumed that the BS knows all RS locations in
RS advance and can obtain UE location in real time. The

= ∫ UE ( ) ∑ (ℎ UR 1 ( UE , ) BS finds routes based on the proposed algorithm
UE

UE and specifies the relay’s beam indices. It is assumed
=1
( ( that an RS is used only for a single route since the
RS
signal multiplexing capability is not considered for

+ ∑ ℎ ( , ) ℎ ( )
UR 2 UE UR R 1 B (18) RSs in this paper. Blockage between RSs and the UE
2
=1
by moving objects, such as vehicle, is ignored. First,
+ ⋯ ∑ ⋯ ∑ ℎ ( , ) in Step I, the relative positions and distances among
⏟ UR UE
−1 the BS, RSs, and UE are calculated according to the
real-time location of UE. It is noted that distances
between the BS and RSs are fixed, while the

∙ ⋯ ℎ ( )) d UE
2
UR R 1 distances between RSs and the UE are varying
quickly. Next, as shown in Step II, a graph of
))
artificial channels is built, in which nodes are BS,
RSs and UE, and weighted edges connect all nodes.
where is the noise signal generated in the RSi. In The distance between two nodes is used as a weight,

Eq.(18), the RS in the first column of Fig. 5 is a noise and the edges of self-loops and NLOS links are
source and is amplified and forwarded in each hop. removed from the graph. Since it is assumed that

48 © International Telecommunication Union, 2021