2026 ASIAN Conference on Communication and Networks (ASIANComNet 2026)

In this paper, we study the issue of resource allo-
cation in semantic-aware vehicular edge computing (VEC). In
the considered setting, computation task of each vehicle may be
performed either locally or in the MEC server by offloading. In
order to reduce transmission overhead, low-dimensional seman-
tics rather than original high-dimensional data of each vehicle
are extracted and transmitted. A problem formulation of the
minimized total delay with respect to energy consumption and
semantics similarity is proposed. Our solution to this problem is
based on deep deterministic policy gradient (DDPG) technique
for optimizing transmit power, extraction ratio, and offloading
decision. The effectiveness of the proposed solution is also
validated by simulation.

semantic communication, offloading decision, VEC, DDPG