The recent worldwide sanitary pandemic has made it clear that changes in user traffic patterns can create load balancing issues in networks (e.g., new peak hours of usage have been observed, especially in suburban residential areas). Such patterns need to be accommodated, often with reliable service quality. Although several studies have examined the user association and resource allocation (UA-RA) issue, there is still no optimal strategy to address such a problem with low complexity while reducing the time overhead. To this end, we propose Performance-Improved Reduced Search Space Simulated Annealing (P IRS3A), an algorithm for solving UA-RA problems in Heterogeneous Networks (HetNets). First, the UA-RA problem is formulated as a multiple 0/1 knapsack problem (MKP) with constraints on the maximum capacity of the base stations (BS) along with the transport block size (TBS) index. Second, the proposed P IRS3A is used to solve the formulated MKP. Simulation results show that P IRS3A outperforms existing schemes in terms of variability and Quality of Service (QoS), including throughput, packet loss ratio (PLR), delay, and jitter. Simulation results also show that P IRS3A generates solutions that are very close to the optimal solution compared to the default simulated annealing (DSA) algorithm.