Information on the transport and phosphorylation rate constants (k1,k2,k3,k4,Ki) of a tracer reflects the biological state of cells. A microfluidic cell culture chip coupled with PSAPD camera (MF-PSAPD) has been developed to give continuous measurements of radioactivity in individual wells. However, constant infusion (CI) of PET tracers through the chambers would give high background activity due to the relatively large volume space of the infusing medium in the wells of cultured cells that compromise the ability of the setup to estimate the k values. New strategies of controlling the infusion and tracer level are needed to provide reliable estimates of the parameters. A switching strategy (SS) was conceived that consists of multiple medium-infusion cycles, each of which has a tracer incubation (TI) period followed by a background-removed (BR) period (tracer-free medium). In this paper, equally switching strategy (ESS) with 12 cycles of constant TI and BR periods (5 min each) was evaluated by computer simulation and by experiments on MF-PSAPD using the tracer fluorodeoxy-glucose (FDG) and the four parameter FDG model. The SS was further optimized by using a simulated annealing algorithm and D-optimal criterion to obtain optimal switching strategy (OSS). Simulations showed that the 12-cycle ESS did not perform as well (i.e., with larger estimated variability of the model parameters) as a 5-cycle OSS that also has multiple practical advantages. Patterns of OSS were found to be insensitive to the variation of cell number and k values, and all tended to have longer TI at the beginning but longer BR at later times. Estimated k values with SS have large reduction in %CV compared to those of CI, with the largest reduction for Ki--from 762% down to 26% under the same count rate conditions. The new optimized strategy of tracer incubation/measurement is able to provide reliable estimates of FDG k's in MF-PSAPD.