Global navigation satellite system performance can be degraded significantly in urban and semi-outdoor environments due to signal blockage and reflection. To enhance smartphone positioning in these areas, integration of pedestrian dead reckoning (PDR) and Bluetooth low energy (BLE) sensors has been proposed, but loosely coupled integration often fails in complex urban environments. This study explores Bluetooth signal characteristics and integrates BLE with GNSS and PDR using three algorithms. GNSS quality is monitored with BLE signals and PDR results to adaptively minimize its impact; smartphone orientation and heading estimated from BLE signals jointly determine walking direction; and a tightly coupled Kalman filter incorporating distance measurements from strong-signal BLE transmitters addresses device heterogeneity and reduces accumulated PDR errors. Experiments on different smartphones show that the proposed system consistently achieves positioning accuracy within 4 m in both urban and semi-outdoor environments.