Page 51 - ITU Journal Future and evolving technologies Volume 3 (2022), Issue 2 – Towards vehicular networks in the 6G era

ITU Journal on Future and Evolving Technologies, Volume 3 (2022), Issue 2

link reception success probability. Therefore, the closest interferer of the receiver j to simplify the
velocity is not included in the physical abstraction problem model. Let us denote the distance between
model in the following discussion. The link success the receiver j and its closest interferer as . Then,

probability is purely defined by the SINR of the link, similar to (10), we can have:
whereas the channel fluctuation is still affected by
the Doppler. , ∝ . (11)

4. PROBLEM FORMULATION Let us refer and as signal distance and main
,

interference distance hereafter. Then, the two
In order to maximize the reliability for the road gradients (10-11) depict the impact of the two
safety in urban scenarios, a broadcast packet is distances on the link reception rate. That is, driving
desired to be received by neighboring vehicles as by the gradient inversely proportional to in (10),

much as possible. While, when one vehicle is the transmitter benefits from reducing the distance
transmitting, the rest of N-1 vehicles in the RoI may from all the receivers. Moreover, before the

decide to transmit or receive in that subframe. Due transmitter moves, if a target vehicle is closer to a
to the half-duplexing operation mode, only the receiver, it will be more beneficial to move even
vehicles in a receiving mode are considered to be closer to that receiver so that the link reception rate
the targeted receivers. Thus, assume there is a or will be maximized. In this way, the node

,
utility associated to each target receiver, i.e., to gains little by moving towards the receivers at the
represent a level of satisfaction that a target vehicle far side, and in a distributed manner the node will
can obtain from successfully receiving a packet. not have the motivation to move towards the far
Then, at the transmitter, its objective will be to nodes to improve their utility. Thus, this is a greedy
move to a better position such that the aggregated utility function without any fairness consideration
utility of its target receivers is maximized. In this between target vehicles.
way, we can formulate node i’s aggregated utility as
follows: As a result, a fair utility function should be
considered to shrink the gain in distance when the
U = ∑ j∈ , (9) is already high. We propose to apply log for the
i
,
,
where stands for the success probability of the in the function (9) and get:
,
,
reception on link from node i to node j and is the log ( ), (12)

set of all receiving nodes in the neighborhood of i. = ∑ j∈ ,
The exact form of function (. ) is yet to be found with the log function, the gradient with respect to
,
via a machine learning approach in Section 5. becomes:
,
However, we will not lose any generality to assume
1 , , (13)
that is a monotonic function of the link , = , ⋅ ,
,
,
defined in (8). Ignoring the influence of Gaussian It means that the increase in terms of in (12) that
noise, the SINR in (8) is directly proportional to the is
signal power of the receiver and inversely can be contributed from an increase in ,
proportional to the sum of the interference caused inversely proportional to the current as well as
,
by the interferers. Thus, we can derive that the its current . The sign of (13) may be the same for
.
,
has a negative relation to , , where , is the different receiving vehicles, which means by getting
distance between the transmitting node i and the closer to any target receiver, it always brings up the
receiving node j. Considering the path loss total utility. The amplitude of the partial derivatives
definition in (3), we have: varies from receiver to receiver. That means, by
moving closer to one target receiver by a unit
, 1 distance, the return in aggregated utility at the
∝ . (10)
transmission node for the whole broadcast packet
, ,
On the contrast, should have a positive relation to is different from what is brought by moving closer

the distance between the receiver j and its to another target receiver. In this way, the
interfering nodes, i.e., transmitting nodes that are transmitter can decide which target receiver is the
close enough (lies in a distance smaller than a pre- most rewarding target to move closer to and hence
defined threshold thI to the receiver j). Although for a gradient-based updating method may be derived
one receiver node j, there could be more than one from there to move the transmitter bit by bit
interferer, it could be meaningful to focus on the iteratively to its optimal position.

© International Telecommunication Union, 2022 39