A constrained optimisation approach for null broadening adaptive beamforming is proposed. This approach improves the robustness of the traditional MVDR beamformer by broadening nulls for interference direction and the mainlobe for the desired direction. This optimisation is efficiently solved by semidefinite programming. The proposed approach, when applied to high altitude platform (HAP) communications using a vertical linear antenna array, provides significantly better coverage performance than a previously reported null broadening technique.Introduction: Antenna beampattern optimisation is widely used to improve the system capacity in wireless communications. Adaptive beamformers are well known for their high resolution and sidelobe suppression [1]. Among them, the Minimum Variance Distortionless Response (MVDR) beamformer [2] is one of the most traditional techniques. However, the performance of the MVDR beamformer significantly degrades if the steering information is inaccurate [3]. The most common technique to improve the robustness of the MVDR beamformer is the diagonal loading [3], [4]. The drawback is that there is currently no reliable method to select the optimal loading factor [5]. Null broadening [6], [7] is another robust approach. In [6], a cluster of equal-strength incoherent sources are artificially distributed around each original source in order to generate a trough like pattern. However, the mainlobe is not broadened. This may result in a poor coverage performance when the system is suffering high steering errors while the mainlobe of the beampattern is narrow. Furthermore, arranging equal number of additive sources for each interferer evidently limits the coverage performance.In this letter, we propose an improved null broadening method. In particular, the following refinements are introduced to improve the robustness of null broadening techniques: 1) a new constrained optimisation problem is formulated to broaden both nulls and mainlobe; 2) the optimisation problem is solved using semidefinite programming (SDP) [8], [9]; 3) nonequal number of additive sources are assigned to different users to reduce the total number of optimisation parameters, which makes the SDP solution more efficient. The proposed approach is compared with the conventional null broadening technique proposed in [6] for downlink communications from high altitude platforms (HAPs) [10]. The performance improvement due to the modified null broadening approach is significant. Null broadening techniques: MVDR adaptive beamforming places nulls in the direction of interferers whilst assigning unit power to the desired user. The N × 1 MVDR weight vectors for the mth desired user is given by [2]